A System for Acquisition, Processing and Visualization of Image Time Series from Multiple Camera Networks
Full text
da a
A icle
A Sys em o Acquisi ion, P ocessing and
Visualiza ion o Image Time Se ies om Mul iple
Came a Ne wo ks
Cemal Melih Tanis 1,*ID , Mikko Pel oniemi 2, Maiju Linkosalmi 1, Mika Au ela 1,
K is in Bö che 3, Te hikki Manninen 1ID and Ali Nadi A slan 1ID
1Finnish Me eo ological Ins i u e (FMI), 00101 Helsinki, Finland; [email p o ec ed] (M.L.);
[email p o ec ed] (M.A.); T[email p o ec ed] (T.M.); Ali.Nadi [email p o ec ed] (A.N.A.)
2Na u al Resou ces Ins i u e Finland (LUKE), 00790 Helsinki, Finland; Mikko.Pel oniem[email p o ec ed]
3Finnish En i onmen Ins i u e (SYKE), 00251 Helsinki, Finland; K is in.Bo che @ympa is o. i
*Co espondence: Cemal.Melih.T[email p o ec ed]
Recei ed: 13 May 2018; Accep ed: 20 June 2018; Published: 24 June 2018
Abs ac :
A sys em o mul iple came a ne wo ks is p oposed o con inuous moni o ing o
ecosys ems by p ocessing image ime se ies. The sys em is buil a ound he Finnish Me eo ological
Image PROcessing Toolbox (FMIPROT), which includes da a acquisi ion, p ocessing and isualiza ion
om mul iple came a ne wo ks. The oolbox has a use - iendly g aphical use in e ace (GUI) o
which only minimal compu e knowledge and skills a e equi ed o use i . Images om came a
ne wo ks a e acqui ed and handled au oma ically acco ding o he common communica ion p o ocols,
e.g., File T ans e P o ocol (FTP). P ocessing ea u es include GUI based selec ion o he egion o
in e es (ROI), au oma ic analysis chain, ex ac ion o ROI based indices such as he g een ac ion
index (GF), ed ac ion index (RF), blue ac ion index (BF), g een- ed ege a ion index (GRVI),
and g een excess (GEI) index, as well as a cus om index de ined by a use -p o ided ma hema ical
o mula. Analysis esul s a e isualized on in e ac i e plo s bo h on he GUI and hype ex ma kup
language (HTML) epo s. The use s can implemen hei own de eloped algo i hms o ex ac
in o ma ion om digi al image se ies o any pu pose. The oolbox can also be un in non-GUI
mode, which allows unning se ies o analyses in se e s una ended and scheduled. The sys em is
demons a ed using an en i onmen al came a ne wo k in Finland.
Keywo ds:
came a ne wo k; digi al image p ocessing; au oma ed p ocessing: use - iendly;
ege a ion phenology; snow co e ; webcam moni o ing; ope a ional moni o ing; da a acquisi ion;
isualiza ion
1. In oduc ion
Webcam pho og aphy is an es ablished way o moni o ing he condi ions in he en i onmen ,
o ins ance o ec ea ion pu poses (ski eso s, hiking ails). Howe e , webcam pho og aphy also
has g owing usage o scien i ic pu poses; i can be applied o con inuously moni o ecosys ems [1].
S udies show ha a ious phenomena o ege a ion phenology can be analyzed by digi al
image y. Di e en ege a ion indices can be ex ac ed o di e en geog aphical egions om digi al
images. Fo di e en ecosys ems and illumina ion condi ions, he indices a y in hei pe o mance on
cha ac e izing phenology e en s [
1
–
13
]. Co ela ion o he ege a ion indices wi h o he phenological
da a om o he senso s has been s udied [
1
,
3
,
5
–
7
,
11
,
14
]. Usage o ege a ion indices o s udy ecological
indica o s, e.g., he nu i ion o wild animals, has also been in es iga ed [
4
,
9
,
15
,
16
]. The iming o
seasonal changes is also analyzed om digi al images [
1
,
10
,
13
,
17
]. Ano he opic o which digi al
Da a 2018,3, 23; doi:10.3390/da a3030023 www.mdpi.com/jou nal/da a
Da a 2018,3, 23 2 o 19
image p ocessing can be used is hyd ology. De ec ion o he amoun o snow co e om digi al images
is s udied o di e en en i onmen s, e.g., moun ains, glacie s [18–22].
The es ablishmen o webcam ne wo ks in ended o scien i ic moni o ing o ecosys ems is
g owing la ge . The Eu opean Phenology Came a Ne wo k (EUROPhen) is a collec ion o came as
used o phenology ac oss Eu ope [
1
]. In a simila way, he PhenoCam Ne wo k has came as om
133 si es ac oss No h Ame ica and Eu ope [
23
]. Time-lapse digi al came a images o Aus alian
biomes o di e en loca ions a e a chi ed and dis ibu ed by he Aus alian Phenocam Ne wo k [
24
].
As a pa o “Clima e Change Indica o s and Vulne abili y o Bo eal Zone Applying Inno a i e
Obse a ion and Modeling Techniques (MONIMET)” p ojec , he MONIMET came a ne wo k
was es ablished o p o ide ime-se ies o ield obse a ion and consis s o abou 30 came as o e
Finland [
25
]. The Phenological Eyes Ne wo k (PEN) is ano he phenological came a ne wo k in
Asia [
26
]. Digi al images and o he phenological da a om such came a ne wo ks ha e been used in
a ious s udies [1,5,10,18,26].
Besides he ne wo ks pa icula ly in ended o scien i ic use, o he ypes o came a ne wo ks
such as a ic webcams, e c. can also be used o scien i ic pu poses. Usage o a ic webcams o
phenological s udies was in es iga ed by Mo is e al. [
27
]. Came a ne wo ks o ski acks and
a alanche moni o ing o e moun ains can also be used in he same manne . These ne wo ks p o ide
po en ial da a o hyd ological pu poses, e.g., analysis o snow co e [
18
,
22
]. On a la ge scale,
he s a us o webcams o e he globe has also been s udied. F eely a ailable images om he global
ne wo k o housands o ou doo webcams ha e g ea po en ial o be used o scien i ic measu emen s.
On he o he hand, he ask o ob aining and o ganizing hese images is signi ican ly challenging [
28
].
Nume ous eely a ailable so wa e packages and ools ha e been de eloped o be used in
analyzing images, bu ew ocus on phenological moni o ing [
21
,
22
,
29
,
30
]. The Phenopix R package is
one example o a package wi h ad anced ea u es o image-based ege a ion phenology [
29
]. I is
designed o compu e ege a ion indices om digi al images. The compu a ions can be done ei he
based on ROI a e ages o pixel alues. Fo ROI-a e aged analyses, he use is able o de ine ROIs
using he GUI dedica ed o ha op ion. The package implemen s se e al da a il e ing and pos
p ocessing me hods. These include di e en cu e i ing me hods, di e en da a il e ing me hods,
ex ac ion o phenophases and unce ain y analysis. The package also o e s unc ions o isualize
and sa e esul s. O he so wa e packages also ha e ea u es simila o Phenopix and o he s a e
di e en [
21
,
22
,
30
]. On he o he hand, exis ing so wa e lacks he ea u es equi ed o examine and
bene i ully om exis ing and u u e came a ne wo ks. Fo achie ing such goals, i is e y impo an
o es ablish sys ems, pla o ms and easy- o-use oolboxes o acqui ing and handling images om
di e en ne wo ks ha ing di e en p o ocols and s anda ds. Conside ing cu en and possible ea u es,
no only o analyzing images o comp ehensi e came a ne wo ks, bu also o c ea ing ope a ional
da a ex ac ion and p ocessing wi h an easy o use oolbox, a mo e cus omized app oach would be o
high alue. This app oach is ollowed in FMIPROT. I is designed o be used as a basis o an au oma ed
p ocessing sys em. This sys em is able o acqui e and p ocess images om mul iple came a ne wo ks
and isualize he p ocessing esul s. I allows o s o ing all he con igu a ions and he op ions used
o analyses, so ha he sys em can be used o c ea e and main ain ope a ional moni o ing se ices.
The plugin sys em allows he use o add new p ocessing algo i hms o FMIPROT, which makes
i possible o bene i om ad anced il e ing, p ocessing and pos -p ocessing me hods in exis ing
so wa e packages.
The sys em a chi ec u e and i s elemen s a e desc ibed in Sec ion 2. In Sec ion 3, mul iple examples
o applica ions which can be used in phenology s udies a e p esen ed. In Sec ion 4, he sys em is
demons a ed as an example o po en ial ope a ional moni o ing using he MONIMET came a
ne wo k. In Sec ion 5, he cu en sys em and he planned ac i i ies a e discussed.
Da a 2018,3, 23 3 o 19
2. Sys em A chi ec u e
The image p ocessing sys em p esen ed in his pape consis s o he Finnish Me eo ological Image
PROcessing Toolbox (FMIPROT), came a ne wo k in o ma ion iles (CNIF), local image eposi o y
and came a ne wo ks which include came as, image eposi o ies and online CNIFs. CNIFs a e he
iles which s o e in o ma ion on how o communica e wi h he came a ne wo ks. The whole sys em
a chi ec u e is illus a ed in Figu e 1and hey a e desc ibed in de ail in he ollowing sec ions.
Da a 2018, 3, x 3 o 19
eposi o y and came a ne wo ks which include came as, image eposi o ies and online CNIFs. CNIFs
a e he iles which s o e in o ma ion on how o communica e wi h he came a ne wo ks. The whole
sys em a chi ec u e is illus a ed in Figu e 1 and hey a e desc ibed in de ail in he ollowing sec ions.
Figu e 1. A chi ec u e o he image p ocessing sys em.
The FMIPROT uns he en i e sys em. The wo k low, including all he ea u es ha a e a ailable
wi h he GUI is explained in he oolbox subsec ion. Those ea u es a e used o s andalone analyses
and de ining pa ame e s o analyses o be used in ope a ional moni o ing. When unning an
ope a ional sys em, he oolbox goes h ough a se ies o asks o each scheduled un. These asks can
be di ided in o h ee phases: (1) acquisi ion; (2) p ocessing; and (3) isualiza ion. I s a s om he
acquisi ion phase by using CNIFs and he empo al selec ion in o ma ion o lis he images om
eposi o ies and download hem i necessa y. In he p ocessing phase, i any olde esul s exis o
he same analysis, he imes amps om he esul s a e ead and lis ed. Using he lis om he olde
esul s, and he ile and imes amp lis s om he acquisi ion phase, he images ha a e no p ocessed,
a e ead and p ocessed acco ding o he p ocessing pa ame e s. The old esul s ha no longe i he
empo al selec ion a e emo ed and he new esul s a e s o ed as aw da a in iles. In he isualiza ion
phase, an HTML epo which includes empo al selec ion, p ocessing pa ame e s, hype links o he
aw da a and in e ac i e plo s ha isualize he esul s. This wo king low o he p oposed
ope a ional sys em wi h all asks in he h ee phases is illus a ed in Figu e 2.
Figu e 2. Wo king low o he p oposed ope a ional sys em.
2.1. Came a Ne wo ks
A came a ne wo k can consis o one o mul iple came as. I is assumed o p o ide ime se ies
o images and an in as uc u e o ecep ion and s o age o hese images. A came a ne wo k
in o ma ion ile (CNIF) is de ined o each came a ne wo k used in FMIPROT. The CNIF includes he
Figu e 1. A chi ec u e o he image p ocessing sys em.
The FMIPROT uns he en i e sys em. The wo k low, including all he ea u es ha a e a ailable
wi h he GUI is explained in he oolbox subsec ion. Those ea u es a e used o s andalone analyses and
de ining pa ame e s o analyses o be used in ope a ional moni o ing. When unning an ope a ional
sys em, he oolbox goes h ough a se ies o asks o each scheduled un. These asks can be di ided
in o h ee phases: (1) acquisi ion; (2) p ocessing; and (3) isualiza ion. I s a s om he acquisi ion
phase by using CNIFs and he empo al selec ion in o ma ion o lis he images om eposi o ies and
download hem i necessa y. In he p ocessing phase, i any olde esul s exis o he same analysis,
he imes amps om he esul s a e ead and lis ed. Using he lis om he olde esul s, and he
ile and imes amp lis s om he acquisi ion phase, he images ha a e no p ocessed, a e ead and
p ocessed acco ding o he p ocessing pa ame e s. The old esul s ha no longe i he empo al
selec ion a e emo ed and he new esul s a e s o ed as aw da a in iles. In he isualiza ion phase,
an HTML epo which includes empo al selec ion, p ocessing pa ame e s, hype links o he aw da a
and in e ac i e plo s ha isualize he esul s. This wo king low o he p oposed ope a ional sys em
wi h all asks in he h ee phases is illus a ed in Figu e 2.
Da a 2018, 3, x 3 o 19
eposi o y and came a ne wo ks which include came as, image eposi o ies and online CNIFs. CNIFs
a e he iles which s o e in o ma ion on how o communica e wi h he came a ne wo ks. The whole
sys em a chi ec u e is illus a ed in Figu e 1 and hey a e desc ibed in de ail in he ollowing sec ions.
Figu e 1. A chi ec u e o he image p ocessing sys em.
The FMIPROT uns he en i e sys em. The wo k low, including all he ea u es ha a e a ailable
wi h he GUI is explained in he oolbox subsec ion. Those ea u es a e used o s andalone analyses
and de ining pa ame e s o analyses o be used in ope a ional moni o ing. When unning an
ope a ional sys em, he oolbox goes h ough a se ies o asks o each scheduled un. These asks can
be di ided in o h ee phases: (1) acquisi ion; (2) p ocessing; and (3) isualiza ion. I s a s om he
acquisi ion phase by using CNIFs and he empo al selec ion in o ma ion o lis he images om
eposi o ies and download hem i necessa y. In he p ocessing phase, i any olde esul s exis o
he same analysis, he imes amps om he esul s a e ead and lis ed. Using he lis om he olde
esul s, and he ile and imes amp lis s om he acquisi ion phase, he images ha a e no p ocessed,
a e ead and p ocessed acco ding o he p ocessing pa ame e s. The old esul s ha no longe i he
empo al selec ion a e emo ed and he new esul s a e s o ed as aw da a in iles. In he isualiza ion
phase, an HTML epo which includes empo al selec ion, p ocessing pa ame e s, hype links o he
aw da a and in e ac i e plo s ha isualize he esul s. This wo king low o he p oposed
ope a ional sys em wi h all asks in he h ee phases is illus a ed in Figu e 2.
Figu e 2. Wo king low o he p oposed ope a ional sys em.
2.1. Came a Ne wo ks
A came a ne wo k can consis o one o mul iple came as. I is assumed o p o ide ime se ies
o images and an in as uc u e o ecep ion and s o age o hese images. A came a ne wo k
in o ma ion ile (CNIF) is de ined o each came a ne wo k used in FMIPROT. The CNIF includes he
Figu e 2. Wo king low o he p oposed ope a ional sys em.
Da a 2018,3, 23 4 o 19
2.1. Came a Ne wo ks
A came a ne wo k can consis o one o mul iple came as. I is assumed o p o ide ime
se ies o images and an in as uc u e o ecep ion and s o age o hese images. A came a ne wo k
in o ma ion ile (CNIF) is de ined o each came a ne wo k used in FMIPROT. The CNIF includes he
equi ed in o ma ion o communica e wi h he came a ne wo k, e.g., communica ion p o ocol, add ess,
ilename o ma . The CNIF can be c ea ed au oma ically using he GUI o manually by using echnical
in o ma ion in he use manual. The CNIF can be s o ed on a local disk o in a emo e se e accessible
by FTP o HTTP. This ea u e allows an ope a o o he came a ne wo k o main ain online CNIF(s) so
ha use s can connec o he came a ne wo k and p ocess he images using FMIPROT.
FMIPROT can be u ilized wi h a ious suppo ed p o ocols and con igu a ions o mul iple
came a ne wo ks. The e is no es ic ion o use ce ain communica ion p o ocols oge he . In Figu e 3,
an example o using FTP o bo h CNIF and a ious came as is illus a ed. Ano he example is gi en
in Figu e 4, using FTP o a ious came as and HTTP o e a web se ice o he came a ne wo k
in o ma ion ile. The MONIMET Came a Ne wo k, which consis s o 28 came as (26 ope a ing)
p oducing 30 image se ies om 14 di e en loca ions in Finland wo ks wi h he same con igu a ion
gi en in Figu e 4[25].
Da a 2018, 3, x 4 o 19
equi ed in o ma ion o communica e wi h he came a ne wo k, e.g., communica ion p o ocol,
add ess, ilename o ma . The CNIF can be c ea ed au oma ically using he GUI o manually by using
echnical in o ma ion in he use manual. The CNIF can be s o ed on a local disk o in a emo e se e
accessible by FTP o HTTP. This ea u e allows an ope a o o he came a ne wo k o main ain online
CNIF(s) so ha use s can connec o he came a ne wo k and p ocess he images using FMIPROT.
FMIPROT can be u ilized wi h a ious suppo ed p o ocols and con igu a ions o mul iple
came a ne wo ks. The e is no es ic ion o use ce ain communica ion p o ocols oge he . In Figu e
3, an example o using FTP o bo h CNIF and a ious came as is illus a ed. Ano he example is
gi en in Figu e 4, using FTP o a ious came as and HTTP o e a web se ice o he came a ne wo k
in o ma ion ile. The MONIMET Came a Ne wo k, which consis s o 28 came as (26 ope a ing)
p oducing 30 image se ies om 14 di e en loca ions in Finland wo ks wi h he same con igu a ion
gi en in Figu e 4 [25].
Figu e 3. A came a ne wo k con igu a ion using File T ans e P o ocol (FTP) o bo h he came a
ne wo k in o ma ion ile and images om a ious came as.
Figu e 4. A came a ne wo k con igu a ion using FTP o images om a ious came as and HTTP
o e a web se ice o he came a ne wo k in o ma ion ile.
FMIPROT can also be connec ed o mul iple came a ne wo ks. Images om each con igu ed
came a ne wo k can be used wi h FMIPROT. A chi ed images om local disk can also be used. In
his case, hese a chi es should be de ined as a sepa a e came a ne wo k. In Figu e 5, an example o
using eigh di e en came a ne wo ks wi h di e en con igu a ions is illus a ed wi h
communica ion p o ocols.
Figu e 3.
A came a ne wo k con igu a ion using File T ans e P o ocol (FTP) o bo h he came a
ne wo k in o ma ion ile and images om a ious came as.
Da a 2018, 3, x 4 o 19
equi ed in o ma ion o communica e wi h he came a ne wo k, e.g., communica ion p o ocol,
add ess, ilename o ma . The CNIF can be c ea ed au oma ically using he GUI o manually by using
echnical in o ma ion in he use manual. The CNIF can be s o ed on a local disk o in a emo e se e
accessible by FTP o HTTP. This ea u e allows an ope a o o he came a ne wo k o main ain online
CNIF(s) so ha use s can connec o he came a ne wo k and p ocess he images using FMIPROT.
FMIPROT can be u ilized wi h a ious suppo ed p o ocols and con igu a ions o mul iple
came a ne wo ks. The e is no es ic ion o use ce ain communica ion p o ocols oge he . In Figu e
3, an example o using FTP o bo h CNIF and a ious came as is illus a ed. Ano he example is
gi en in Figu e 4, using FTP o a ious came as and HTTP o e a web se ice o he came a ne wo k
in o ma ion ile. The MONIMET Came a Ne wo k, which consis s o 28 came as (26 ope a ing)
p oducing 30 image se ies om 14 di e en loca ions in Finland wo ks wi h he same con igu a ion
gi en in Figu e 4 [25].
Figu e 3. A came a ne wo k con igu a ion using File T ans e P o ocol (FTP) o bo h he came a
ne wo k in o ma ion ile and images om a ious came as.
Figu e 4. A came a ne wo k con igu a ion using FTP o images om a ious came as and HTTP
o e a web se ice o he came a ne wo k in o ma ion ile.
FMIPROT can also be connec ed o mul iple came a ne wo ks. Images om each con igu ed
came a ne wo k can be used wi h FMIPROT. A chi ed images om local disk can also be used. In
his case, hese a chi es should be de ined as a sepa a e came a ne wo k. In Figu e 5, an example o
using eigh di e en came a ne wo ks wi h di e en con igu a ions is illus a ed wi h
communica ion p o ocols.
Figu e 4.
A came a ne wo k con igu a ion using FTP o images om a ious came as and HTTP o e
a web se ice o he came a ne wo k in o ma ion ile.
FMIPROT can also be connec ed o mul iple came a ne wo ks. Images om each con igu ed
came a ne wo k can be used wi h FMIPROT. A chi ed images om local disk can also be used.
In his case, hese a chi es should be de ined as a sepa a e came a ne wo k. In Figu e 5,
an example o using eigh di e en came a ne wo ks wi h di e en con igu a ions is illus a ed
wi h communica ion p o ocols.
Da a 2018,3, 23 5 o 19
Da a 2018, 3, x 5 o 19
Figu e 5. An example o using eigh di e en came a ne wo ks wi h di e en con igu a ions.
2.2. FMIPROT
2.2.1. Fea u es
The main ea u es o he FMIPROT a e summa ized below. All he de ailed in o ma ion abou
FMIPROT and he ins uc ions can be ound in he use manual [31].
The oolbox has a s aigh o wa d ins alla ion, which equi es minimal compu e skills. I does
no ha e any so wa e dependencies ha equi e any ex a ins alla ions (excep old Linux sys ems).
The in e ace is compac and use - iendly. All he ac ions equi ed o c ea e and un analyses can be
pe o med using he GUI. I also has in e ac i e in e aces whe e necessa y, o example, o selec
ROIs. In Figu e 6, an example o ROI selec ion (a) and an example o plo ing o esul s (b) a e
shown. The analyses ha a e se up using GUI can also be un h ough he command line in e ace
(CLI). I is possible o use he sys em o ope a ional moni o ing by using such ea u es on emo e
se e s.
FMIPROT p o ides au oma ic downloading and handling o images om mul iple came a
ne wo ks which can be added o he oolbox using he GUI. Use s selec he sou ce(s) o images, i.e.,
he came a(s), and he so wa e downloads he images om he came a ne wo k se e s. The
downloaded images a e s o ed in a local di ec o y used in he analyses be o e being cas o he
p ocessing unc ions. I is also possible o use p oxy pa ame e s o download he images om came a
ne wo ks. Fo example, i he image eposi o y o a came a is hos ed on an FTP se e , bu he
eposi o y is also accessible as a di ec o y on local disk, a “came a ne wo k p oxy” can be used o
ead images om he disk di ec ly ins ead o connec ing o he FTP se e . Simila ly, di e en
c eden ials o he FTP connec ion can be used. A use can s o e a se up ile which includes only he
use name o he FTP connec ion o he image eposi o y, which is only accessible by ano he use
wi h he co ec use name and passwo d. In ha case, he oolbox will use he p oxy o eplace he
c eden ials o he connec ion.
Figu e 5. An example o using eigh di e en came a ne wo ks wi h di e en con igu a ions.
2.2. FMIPROT
2.2.1. Fea u es
The main ea u es o he FMIPROT a e summa ized below. All he de ailed in o ma ion abou
FMIPROT and he ins uc ions can be ound in he use manual [31].
The oolbox has a s aigh o wa d ins alla ion, which equi es minimal compu e skills. I does
no ha e any so wa e dependencies ha equi e any ex a ins alla ions (excep old Linux sys ems).
The in e ace is compac and use - iendly. All he ac ions equi ed o c ea e and un analyses can
be pe o med using he GUI. I also has in e ac i e in e aces whe e necessa y, o example, o selec
ROIs. In Figu e 6, an example o ROI selec ion (a) and an example o plo ing o esul s (b) a e shown.
The analyses ha a e se up using GUI can also be un h ough he command line in e ace (CLI). I is
possible o use he sys em o ope a ional moni o ing by using such ea u es on emo e se e s.
FMIPROT p o ides au oma ic downloading and handling o images om mul iple came a
ne wo ks which can be added o he oolbox using he GUI. Use s selec he sou ce(s) o images,
i.e., he came a(s)
, and he so wa e downloads he images om he came a ne wo k se e s.
The downloaded images a e s o ed in a local di ec o y used in he analyses be o e being cas o
he p ocessing unc ions. I is also possible o use p oxy pa ame e s o download he images om
came a ne wo ks. Fo example, i he image eposi o y o a came a is hos ed on an FTP se e , bu he
eposi o y is also accessible as a di ec o y on local disk, a “came a ne wo k p oxy” can be used o ead
images om he disk di ec ly ins ead o connec ing o he FTP se e . Simila ly, di e en c eden ials
o he FTP connec ion can be used. A use can s o e a se up ile which includes only he use name o
he FTP connec ion o he image eposi o y, which is only accessible by ano he use wi h he co ec
use name and passwo d. In ha case, he oolbox will use he p oxy o eplace he c eden ials o
he connec ion.
Da a 2018,3, 23 6 o 19
Da a 2018, 3, x 6 o 19
(a)
(b)
Figu e 6. Finnish Me eo ological Image PROcessing Toolbox (FMIPROT) G aphical Use In e ace
examples: egion o in e es (ROI) selec ion (a); plo ing o esul s (b). Two examples a e no ela ed
o each o he , i.e., he esul s in (b) a e no om an analysis using he ROIs in (a).
FMIPROT is designed so ha he use can wo k wi h mul iple analyses a he same ime. The
op ions and pa ame e s o hese mul iple analyses a e se up in a hie a chical way. Fo each un o
he so wa e, he use has a “Se up” which may ha e mul iple “Scena ios”. These scena ios may also
ha e mul iple “Analyses”. This hie a chy allows he use o manage and un all analyses needed o
a possible s udy o an ope a ional p oduc . An analysis is an algo i hm which is applied o selec ed
images. Fo example, calcula ing he “G eenness Index” wi h a speci ic algo i hm is an analysis. A
scena io is a combina ion o op ions ha a e se o one o mul iple analyses o be done o images
om a single came a. A se up is a collec ion o scena ios. Se ups can be sa ed as an “FMIPROT Se up
File”. Se up iles can be loaded o he so wa e o es o ing all op ions. This ea u e makes i possible
o sha e op ions o analyses in a mo e agile way.
Scena ios include op ions such as came a selec ion, empo al selec ion o images, h eshold
selec ions o il e ing pixels and images, masking pa ame e s o selec ing ROIs and p ocessing
pa ame e s o analyses.
A he end o unning analyses, he oolbox c ea es a “se up epo ” which includes all he
analysis op ions o each scena io and he esul s o each analysis on in e ac i e plo s. The epo is
in HTML o ma and uses Ja aSc ip , hus i can be opened by any mode n b owse . In he epo ,
Figu e 6.
Finnish Me eo ological Image PROcessing Toolbox (FMIPROT) G aphical Use In e ace
examples: egion o in e es (ROI) selec ion (
a
); plo ing o esul s (
b
). Two examples a e no ela ed o
each o he , i.e., he esul s in (b) a e no om an analysis using he ROIs in (a).
FMIPROT is designed so ha he use can wo k wi h mul iple analyses a he same ime.
The op ions and pa ame e s o hese mul iple analyses a e se up in a hie a chical way. Fo each
un o he so wa e, he use has a “Se up” which may ha e mul iple “Scena ios”. These scena ios
may also ha e mul iple “Analyses”. This hie a chy allows he use o manage and un all analyses
needed o a possible s udy o an ope a ional p oduc . An analysis is an algo i hm which is applied
o selec ed images. Fo example, calcula ing he “G eenness Index” wi h a speci ic algo i hm is an
analysis. A scena io is a combina ion o op ions ha a e se o one o mul iple analyses o be done
o images om a single came a. A se up is a collec ion o scena ios. Se ups can be sa ed as an
“FMIPROT Se up File”. Se up iles can be loaded o he so wa e o es o ing all op ions. This ea u e
makes i possible o sha e op ions o analyses in a mo e agile way.
Da a 2018,3, 23 7 o 19
Scena ios include op ions such as came a selec ion, empo al selec ion o images,
h eshold selec ions o il e ing pixels and images, masking pa ame e s o selec ing ROIs and
p ocessing pa ame e s o analyses.
A he end o unning analyses, he oolbox c ea es a “se up epo ” which includes all he analysis
op ions o each scena io and he esul s o each analysis on in e ac i e plo s. The epo is in HTML
o ma and uses Ja aSc ip , hus i can be opened by any mode n b owse . In he epo , he e a e also
links o he da a iles o he esul s. When he epo is au oma ically opened in he de aul b owse
a e he analyses, use s can each he da a iles by hese links. “Se up epo s” can also be c ea ed
wi hou any esul s by using he op ion om he “se up” menu. The esul s can also be p e iewed
in he GUI. The plo s in he GUI can also be cus omized in nume ous ways such as a ying colo s,
lines and ma ke s. The e is also an op ion o sa e plo s as image iles.
FMIPROT has a plugin sys em, which allows use s o add hei own algo i hms o he oolbox. I is
planned o g adually implemen a la ge a ie y o digi al image p ocessing algo i hms o he FMIPROT.
This ea u e allows use s o plug in hei own algo i hms using ad anced il e ing, p ocessing and
pos -p ocessing me hods in exis ing so wa e packages while ge ing he bene i s o he o he exis ing
ea u es o he FMIPROT.
The so wa e is able o check he empo al s a us o he image se ies o came a ne wo ks in a
quan i a i e way. When he analysis is un, i scans all he images in he image eposi o y and c ea es
esul iles which include in o ma ion abou he numbe o images a ime se ies. The esul iles can
also be plo ed o isualize he empo al co e age.
2.2.2. So wa e
FMIPROT is de eloped in Py hon p og amming language, using nume ous eely a ailable
modules. The Py hon language is chosen o he ease o debugging and simplici y. I o e s he
possibili y o un he code wi hou compiling manually. The Py hon code is w i en in a modula way
o inc ease eadabili y and compa ibili y o possible upda es and in eg a ing codes w i en in o he
languages. The “Maho as” package is used o image p ocessing [
32
]. Tkin e is used o he g aphical
use in e ace.
The oolbox is a ailable o bo h Linux and Windows pla o ms. The so wa e packages can be
downloaded om he FMIPROT websi e, a h p:// mip o . mi. i. They also include a de ailed use
manual wi h explana ions on ins alla ion, usage and cu en algo i hms used in he oolbox.
2.2.3. Wo k low
On he code le el, wo k low o he p og am has many s eps wi h complex ela ions due o
he g aphical use in e ace. Fo he sake o he p ocessing sys em, he wo k low is explained in a
s aigh o wa d way by dis ega ding code le el de ails.
When he so wa e is s a ed in GUI mode, i i s ini ializes he use in e ace and eads he
communica ion pa ame e s o all came a ne wo ks. When i is eady, he use in e ace wai s o he
inpu s o be p ocessed. Acco ding o he inpu s o he GUI, he so wa e gi es esponses. A e each
esponse, he use in e ace wai s o ano he inpu . Scena io op ions and se ings a e modi ied
acco ding o hose inpu s and changes a e applied immedia ely. I eques ed, scena io op ions a e
s o ed on iles. When he “Run” inpu is gi en, i s a s he p ocessing chain o he analysis o he
images. Du ing he p ocessing, he in e ace does no espond o he inpu s. I comple es he necessa y
s eps one by one, acco ding o he scena io op ions and p og am se ings. Those s eps can be seen
in Figu e 7. A e he p ocessing chain, he in e ace opens bo h he “Resul Viewe ” menu and he
“Se up epo ” o isualize he esul s. Finally, he use in e ace becomes esponsi e again o u he
in e ac ion. I he p og am is un in non-GUI mode, he p og am skips he GUI p omp and di ec ly
uns he gi en command.
Da a 2018,3, 23 8 o 19
Da a 2018, 3, x 8 o 19
Figu e 7. Simpli ied wo k low o FMIPROT.
2.3. Image P ocessing Algo i hms
2.3.1. Colo F ac ion Indices (Colo Ch oma ic Coo dina es)
The colo ac ion index is he a io o he sum (o mean) o he digi al alues o ha colo in a
selec ed a ea o he sum (o mean) o digi al alues o all he colo channels in ha a ea in an image.
The ed ac ion index (RF) (o al e na i ely ed ch oma ic coo dina e (RCC), s eng h o signal in ed
channel (SR), ed a io, ed index (RI), ela i e b igh ness o ed channel), g een ac ion index (GF)
(o al e na i ely g een ch oma ic coo dina e (GCC), s eng h o signal in g een channel (SG), g een
a io, g een index (GI), ela i e b igh ness o g een channel), blue ac ion index (BF) (o al e na i ely
De ine
Modi y
Qui
Run
S a
Ini ialize GUI
GUI
P omp
Load came a
ne wo ks
Scena ios &
se ings
Lis and/o
download images
Analyze images
Fil e images
S o e esul s
Gene a e epo
Show esul s
End
Resul s
Repo
Figu e 7. Simpli ied wo k low o FMIPROT.
2.3. Image P ocessing Algo i hms
2.3.1. Colo F ac ion Indices (Colo Ch oma ic Coo dina es)
The colo ac ion index is he a io o he sum (o mean) o he digi al alues o ha colo in
a selec ed a ea o he sum (o mean) o digi al alues o all he colo channels in ha a ea in an
image. The ed ac ion index (RF) (o al e na i ely ed ch oma ic coo dina e (RCC), s eng h o
signal in ed channel (S
R
), ed a io, ed index (RI), ela i e b igh ness o ed channel), g een ac ion
index (GF) (o al e na i ely g een ch oma ic coo dina e (GCC), s eng h o signal in g een channel
Da a 2018,3, 23 9 o 19
(S
G
), g een a io, g een index (GI), ela i e b igh ness o g een channel), blue ac ion index (BF)
(o al e na i ely blue ch oma ic coo dina e (BCC), s eng h o signal in blue channel (S
B
), blue a io,
blue index (BI), ela i e b igh ness o blue channel), b igh ness (BR) and luminance (L) a e calcula ed as;
RF =R/(R+G+B) (1)
GF =G/(R+G+B) (2)
BF =B/(R+G+B) (3)
BR =R+G+B(4)
L= 0.2989 R+ 0.5870 G+ 0.1140 B(5)
whe e R,Gand Ba e he digi al numbe s o he ed, g een and blue channel [
7
,
11
,
13
,
14
].
B igh ness and luminance alues a e no malized o alues be ween 0 and 1 in he analysis.
Time se ies o colo ac ion indices om webcam images can be used in he de e mina ion
o phenological e en s. Changes in ege a ion du ing seasons can be moni o ed using GF and RF,
by de ec ing he da es ha hese changes occu [
11
,
13
]. The da es can be de ec ed in pos -p ocessing
by i ing GF and RF da a which is p oduced by FMIPROT o sigmoid cu es and inding he ime o
he as es change by using second o de de i a i es o he i ed cu es. I is planned o implemen
he ex ac ion o phenophases by de eloping a plugin o use cu e i ing me hods om he Phenopix
Rpackage.
2.3.2. G een-Red Vege a ion Index (GRVI)
The GRVI uses he di e ence be ween g een and ed colo s. Because o he nega i e ed e m,
GRVI can ge nega i e o posi i e alues depending on he season and a ge . Du ing he g owing
season GRVI should esul in posi i e alues and du ing he do man season i should esul in nega i e
alues o deciduous species. The GRVI o an image is calcula ed as,
GRVI = (G−R)/(G+R) (6)
whe e Rand Ga e digi al numbe alues o ed and g een channels [6–8].
In e ms o calcula ion o he index, GRVI is simila o he widely used NDVI, whe e nea -in a ed
e lec ance is used ins ead o g een e lec ance. The di e ences be ween hese wo indices a e discussed
by Mo ohka e al., whe e he applicabili y o GRVI o ege a ion phenology is s udied. One bene i o
using GRVI is ha i is be e sui ed o de ec ing a sub le dis u bance in he middle o he g owing
pe iod [
8
]. Ano he ad an age is ha GRVI esul s in di e en pa e ns o seasonal change o
deciduous o es s and g assland.
2.3.3. G een Excess Index (2G- bi)
The g een excess index uses he di e ence be ween bo h g een- ed and g een-blue. The GEI o an
image is calcula ed as,
GEI = 2G−(R+B) (7)
whe e R,Gand Ba e digi al numbe alues o ed, g een and blue channels [
11
]. The mean GEI o
ROIs is calcula ed by including pixels o which none o he channel digi al alues a e ze o.
2.3.4. Cus om Colo Index
This algo i hm calcula es an index om a ma hema ical o mula en e ed by he use using a e age
alues o ed, g een and blue channels in ROIs. The o mula suppo s sums, di e ences, mul iplica ion,
and di ision and ope a ion p io i y by using pa en heses. Using his algo i hm, he use can analyze
images using new indices o he indices which a e in he li e a u e bu no implemen ed in he oolbox.
Da a 2018,3, 23 16 o 19
ha includes he esul s. The esul s a e amed in he came a pages on he MONIMET websi e,
e.g., h p://monime . mi. i/?page=Came as&camid=Hyy iala_Pine_C own
. In he came a pages,
he came a can be selec ed and also ope a ional p ocessing esul s o he selec ed came a can be
seen. The plo s in he came a pages also ha e links o he da a and he se up epo , whe e he
scena io op ions can be checked. In Table 1, 4 ROIs om 3 came as a e lis ed wi h p e iew images and
sc eensho s o he plo s in hei HTML epo s. The plo s a e smoo hed by he mo ing a e age me hod
ia he ea u e on he plo in e ace. Fi een da a poin s (3 days o da a) a e used o smoo hing.
Table 1.
P e iew images and sc eensho s o he HTML epo plo s om he demons a ed sys em o
4 ROIs om 3 came as: (1) Bi ch ee om Hyy iälä c own came a; (2) Sp uce ee om Hyy iälä c own
came a; (3) Sp uce ee om Ken ä o a g ound came a and (4) Bi ch ee om Lammi c own came a.
ROI P e iew Image GF Plo
1
Da a 2018, 3, x 16 o 19
Table 1. P e iew images and sc eensho s o he HTML epo plo s om he demons a ed sys em
o 4 ROIs om 3 came as: (1) Bi ch ee om Hyy iälä c own came a; (2) Sp uce ee om Hyy iälä
c own came a; (3) Sp uce ee om Ken ä o a g ound came a and (4) Bi ch ee om Lammi c own
came a.
ROI P e iew Image GF Plo
1
2
3
4
The se up in his demons a ion will con inue e ol es and be main ained. The con igu a ion o
he analyses may change and mo e calcula ions may be added in ime, bu he p inciple will s ay he
same.
5. Discussion and Fu he De elopmen s
We p esen ed a new sys em o mul iple came a ne wo ks and a oolbox, FMIPROT, wi h he
ease o use and applicabili y o analyzing digi al images, and demons a ed i s use o ex ac ing
ege a ion indices ime se ies. In addi ion o implemen ed ege a ion index algo i hms, he
FMIPROT allows use s o implemen hei own de eloped algo i hms. Ano he s ong ea u e is ha
FMIPROT can communica e wi h mul iple came a ne wo ks as shown in Sec ion 2.3. The a chi ec u e
o he sys em shows ha i is sui able o apid u he imp o emen s and addi ional ea u es.
Cu en de elopmen o he FMIPROT includes p ocessing image ime se ies om came a
ne wo ks o ex ac ing ac ional snow co e . The algo i hms o be used o he ex ac ion a e s udied
by A slan e al., including he in eg a ion o he algo i hms o FMIPROT and he esul s o images
om ou came as in MONIMET came a ne wo k [18]. F ac ional snow co e ex ac ion is planned
Da a 2018, 3, x 16 o 19
Table 1. P e iew images and sc eensho s o he HTML epo plo s om he demons a ed sys em
o 4 ROIs om 3 came as: (1) Bi ch ee om Hyy iälä c own came a; (2) Sp uce ee om Hyy iälä
c own came a; (3) Sp uce ee om Ken ä o a g ound came a and (4) Bi ch ee om Lammi c own
came a.
ROI P e iew Image GF Plo
1
2
3
4
The se up in his demons a ion will con inue e ol es and be main ained. The con igu a ion o
he analyses may change and mo e calcula ions may be added in ime, bu he p inciple will s ay he
same.
5. Discussion and Fu he De elopmen s
We p esen ed a new sys em o mul iple came a ne wo ks and a oolbox, FMIPROT, wi h he
ease o use and applicabili y o analyzing digi al images, and demons a ed i s use o ex ac ing
ege a ion indices ime se ies. In addi ion o implemen ed ege a ion index algo i hms, he
FMIPROT allows use s o implemen hei own de eloped algo i hms. Ano he s ong ea u e is ha
FMIPROT can communica e wi h mul iple came a ne wo ks as shown in Sec ion 2.3. The a chi ec u e
o he sys em shows ha i is sui able o apid u he imp o emen s and addi ional ea u es.
Cu en de elopmen o he FMIPROT includes p ocessing image ime se ies om came a
ne wo ks o ex ac ing ac ional snow co e . The algo i hms o be used o he ex ac ion a e s udied
by A slan e al., including he in eg a ion o he algo i hms o FMIPROT and he esul s o images
om ou came as in MONIMET came a ne wo k [18]. F ac ional snow co e ex ac ion is planned
2
Da a 2018, 3, x 16 o 19
Table 1. P e iew images and sc eensho s o he HTML epo plo s om he demons a ed sys em
o 4 ROIs om 3 came as: (1) Bi ch ee om Hyy iälä c own came a; (2) Sp uce ee om Hyy iälä
c own came a; (3) Sp uce ee om Ken ä o a g ound came a and (4) Bi ch ee om Lammi c own
came a.
ROI P e iew Image GF Plo
1
2
3
4
The se up in his demons a ion will con inue e ol es and be main ained. The con igu a ion o
he analyses may change and mo e calcula ions may be added in ime, bu he p inciple will s ay he
same.
5. Discussion and Fu he De elopmen s
We p esen ed a new sys em o mul iple came a ne wo ks and a oolbox, FMIPROT, wi h he
ease o use and applicabili y o analyzing digi al images, and demons a ed i s use o ex ac ing
ege a ion indices ime se ies. In addi ion o implemen ed ege a ion index algo i hms, he
FMIPROT allows use s o implemen hei own de eloped algo i hms. Ano he s ong ea u e is ha
FMIPROT can communica e wi h mul iple came a ne wo ks as shown in Sec ion 2.3. The a chi ec u e
o he sys em shows ha i is sui able o apid u he imp o emen s and addi ional ea u es.
Cu en de elopmen o he FMIPROT includes p ocessing image ime se ies om came a
ne wo ks o ex ac ing ac ional snow co e . The algo i hms o be used o he ex ac ion a e s udied
by A slan e al., including he in eg a ion o he algo i hms o FMIPROT and he esul s o images
om ou came as in MONIMET came a ne wo k [18]. F ac ional snow co e ex ac ion is planned
Da a 2018, 3, x 16 o 19
Table 1. P e iew images and sc eensho s o he HTML epo plo s om he demons a ed sys em
o 4 ROIs om 3 came as: (1) Bi ch ee om Hyy iälä c own came a; (2) Sp uce ee om Hyy iälä
c own came a; (3) Sp uce ee om Ken ä o a g ound came a and (4) Bi ch ee om Lammi c own
came a.
ROI P e iew Image GF Plo
1
2
3
4
The se up in his demons a ion will con inue e ol es and be main ained. The con igu a ion o
he analyses may change and mo e calcula ions may be added in ime, bu he p inciple will s ay he
same.
5. Discussion and Fu he De elopmen s
We p esen ed a new sys em o mul iple came a ne wo ks and a oolbox, FMIPROT, wi h he
ease o use and applicabili y o analyzing digi al images, and demons a ed i s use o ex ac ing
ege a ion indices ime se ies. In addi ion o implemen ed ege a ion index algo i hms, he
FMIPROT allows use s o implemen hei own de eloped algo i hms. Ano he s ong ea u e is ha
FMIPROT can communica e wi h mul iple came a ne wo ks as shown in Sec ion 2.3. The a chi ec u e
o he sys em shows ha i is sui able o apid u he imp o emen s and addi ional ea u es.
Cu en de elopmen o he FMIPROT includes p ocessing image ime se ies om came a
ne wo ks o ex ac ing ac ional snow co e . The algo i hms o be used o he ex ac ion a e s udied
by A slan e al., including he in eg a ion o he algo i hms o FMIPROT and he esul s o images
om ou came as in MONIMET came a ne wo k [18]. F ac ional snow co e ex ac ion is planned
3
Da a 2018, 3, x 16 o 19
Table 1. P e iew images and sc eensho s o he HTML epo plo s om he demons a ed sys em
o 4 ROIs om 3 came as: (1) Bi ch ee om Hyy iälä c own came a; (2) Sp uce ee om Hyy iälä
c own came a; (3) Sp uce ee om Ken ä o a g ound came a and (4) Bi ch ee om Lammi c own
came a.
ROI P e iew Image GF Plo
1
2
3
4
The se up in his demons a ion will con inue e ol es and be main ained. The con igu a ion o
he analyses may change and mo e calcula ions may be added in ime, bu he p inciple will s ay he
same.
5. Discussion and Fu he De elopmen s
We p esen ed a new sys em o mul iple came a ne wo ks and a oolbox, FMIPROT, wi h he
ease o use and applicabili y o analyzing digi al images, and demons a ed i s use o ex ac ing
ege a ion indices ime se ies. In addi ion o implemen ed ege a ion index algo i hms, he
FMIPROT allows use s o implemen hei own de eloped algo i hms. Ano he s ong ea u e is ha
FMIPROT can communica e wi h mul iple came a ne wo ks as shown in Sec ion 2.3. The a chi ec u e
o he sys em shows ha i is sui able o apid u he imp o emen s and addi ional ea u es.
Cu en de elopmen o he FMIPROT includes p ocessing image ime se ies om came a
ne wo ks o ex ac ing ac ional snow co e . The algo i hms o be used o he ex ac ion a e s udied
by A slan e al., including he in eg a ion o he algo i hms o FMIPROT and he esul s o images
om ou came as in MONIMET came a ne wo k [18]. F ac ional snow co e ex ac ion is planned
Da a 2018, 3, x 16 o 19
Table 1. P e iew images and sc eensho s o he HTML epo plo s om he demons a ed sys em
o 4 ROIs om 3 came as: (1) Bi ch ee om Hyy iälä c own came a; (2) Sp uce ee om Hyy iälä
c own came a; (3) Sp uce ee om Ken ä o a g ound came a and (4) Bi ch ee om Lammi c own
came a.
ROI P e iew Image GF Plo
1
2
3
4
The se up in his demons a ion will con inue e ol es and be main ained. The con igu a ion o
he analyses may change and mo e calcula ions may be added in ime, bu he p inciple will s ay he
same.
5. Discussion and Fu he De elopmen s
We p esen ed a new sys em o mul iple came a ne wo ks and a oolbox, FMIPROT, wi h he
ease o use and applicabili y o analyzing digi al images, and demons a ed i s use o ex ac ing
ege a ion indices ime se ies. In addi ion o implemen ed ege a ion index algo i hms, he
FMIPROT allows use s o implemen hei own de eloped algo i hms. Ano he s ong ea u e is ha
FMIPROT can communica e wi h mul iple came a ne wo ks as shown in Sec ion 2.3. The a chi ec u e
o he sys em shows ha i is sui able o apid u he imp o emen s and addi ional ea u es.
Cu en de elopmen o he FMIPROT includes p ocessing image ime se ies om came a
ne wo ks o ex ac ing ac ional snow co e . The algo i hms o be used o he ex ac ion a e s udied
by A slan e al., including he in eg a ion o he algo i hms o FMIPROT and he esul s o images
om ou came as in MONIMET came a ne wo k [18]. F ac ional snow co e ex ac ion is planned
4
Da a 2018, 3, x 16 o 19
Table 1. P e iew images and sc eensho s o he HTML epo plo s om he demons a ed sys em
o 4 ROIs om 3 came as: (1) Bi ch ee om Hyy iälä c own came a; (2) Sp uce ee om Hyy iälä
c own came a; (3) Sp uce ee om Ken ä o a g ound came a and (4) Bi ch ee om Lammi c own
came a.
ROI P e iew Image GF Plo
1
2
3
4
The se up in his demons a ion will con inue e ol es and be main ained. The con igu a ion o
he analyses may change and mo e calcula ions may be added in ime, bu he p inciple will s ay he
same.
5. Discussion and Fu he De elopmen s
We p esen ed a new sys em o mul iple came a ne wo ks and a oolbox, FMIPROT, wi h he
ease o use and applicabili y o analyzing digi al images, and demons a ed i s use o ex ac ing
ege a ion indices ime se ies. In addi ion o implemen ed ege a ion index algo i hms, he
FMIPROT allows use s o implemen hei own de eloped algo i hms. Ano he s ong ea u e is ha
FMIPROT can communica e wi h mul iple came a ne wo ks as shown in Sec ion 2.3. The a chi ec u e
o he sys em shows ha i is sui able o apid u he imp o emen s and addi ional ea u es.
Cu en de elopmen o he FMIPROT includes p ocessing image ime se ies om came a
ne wo ks o ex ac ing ac ional snow co e . The algo i hms o be used o he ex ac ion a e s udied
by A slan e al., including he in eg a ion o he algo i hms o FMIPROT and he esul s o images
om ou came as in MONIMET came a ne wo k [18]. F ac ional snow co e ex ac ion is planned
Da a 2018, 3, x 16 o 19
Table 1. P e iew images and sc eensho s o he HTML epo plo s om he demons a ed sys em
o 4 ROIs om 3 came as: (1) Bi ch ee om Hyy iälä c own came a; (2) Sp uce ee om Hyy iälä
c own came a; (3) Sp uce ee om Ken ä o a g ound came a and (4) Bi ch ee om Lammi c own
came a.
ROI P e iew Image GF Plo
1
2
3
4
The se up in his demons a ion will con inue e ol es and be main ained. The con igu a ion o
he analyses may change and mo e calcula ions may be added in ime, bu he p inciple will s ay he
same.
5. Discussion and Fu he De elopmen s
We p esen ed a new sys em o mul iple came a ne wo ks and a oolbox, FMIPROT, wi h he
ease o use and applicabili y o analyzing digi al images, and demons a ed i s use o ex ac ing
ege a ion indices ime se ies. In addi ion o implemen ed ege a ion index algo i hms, he
FMIPROT allows use s o implemen hei own de eloped algo i hms. Ano he s ong ea u e is ha
FMIPROT can communica e wi h mul iple came a ne wo ks as shown in Sec ion 2.3. The a chi ec u e
o he sys em shows ha i is sui able o apid u he imp o emen s and addi ional ea u es.
Cu en de elopmen o he FMIPROT includes p ocessing image ime se ies om came a
ne wo ks o ex ac ing ac ional snow co e . The algo i hms o be used o he ex ac ion a e s udied
by A slan e al., including he in eg a ion o he algo i hms o FMIPROT and he esul s o images
om ou came as in MONIMET came a ne wo k [18]. F ac ional snow co e ex ac ion is planned
The se up in his demons a ion will con inue e ol es and be main ained. The con igu a ion
o he analyses may change and mo e calcula ions may be added in ime, bu he p inciple will s ay
he same.
5. Discussion and Fu he De elopmen s
We p esen ed a new sys em o mul iple came a ne wo ks and a oolbox, FMIPROT, wi h he
ease o use and applicabili y o analyzing digi al images, and demons a ed i s use o ex ac ing
ege a ion indices ime se ies. In addi ion o implemen ed ege a ion index algo i hms, he FMIPROT
Da a 2018,3, 23 17 o 19
allows use s o implemen hei own de eloped algo i hms. Ano he s ong ea u e is ha FMIPROT
can communica e wi h mul iple came a ne wo ks as shown in Sec ion 2.3. The a chi ec u e o he
sys em shows ha i is sui able o apid u he imp o emen s and addi ional ea u es.
Cu en de elopmen o he FMIPROT includes p ocessing image ime se ies om came a
ne wo ks o ex ac ing ac ional snow co e . The algo i hms o be used o he ex ac ion a e
s udied by A slan e al., including he in eg a ion o he algo i hms o FMIPROT and he esul s
o images om ou came as in MONIMET came a ne wo k [
18
]. F ac ional snow co e ex ac ion
is planned o be included in he FMIPROT wi h he elease o he nex e sion. I is also used in
ongoing s udies and will be used in planned s udies, which include he es ablishmen o ope a ional
moni o ing o snow co e using he came as in he Solid P ecipi a ion In e compa ison Expe imen
(SPICE) si e(s) in Finland and MONIMET came a ne wo k and alida ion o di e en sa elli e-based
snow co e p oduc s, and ope a ional moni o ing o ege a ion phenology wi h MONIMET came a
ne wo k [
33
,
35
]. I is also planned o implemen a web se ice o use s o con inuously upload hei
came a images which can be p ocessed using he same web se ice wi h FMIPROT in he backend.
Fu u e de elopmen s o FMIPROT ha a e planned include (1) o de elop suppo o o he
image ypes, such as ea h obse a ion da a (e.g., Sen inel-2, Sen inel-3, Landsa ) wi h mul iple
channels; (2) o design mo e use - iendly GUI; (3) o implemen some compa ison ools o au oma ic
alida ion by compa ing he da a ex ac ed om webcams o o he da a sou ces; (4) o implemen mo e
algo i hms o analyzing and pos p ocessing digi al images; and (5) scheduled asks o ope a ional
moni o ing. Those ea u es will allow he ex ension o au oma ed p ocessing o images om mul iple
came a ne wo ks o ope a ional da a ex ac ion and alida ion o a ious en i onmen al pa ame e s,
which a e e y impo an o di e en applica ions, including:
•Moni o ing land co e change o en i onmen al moni o ing
•Ag icul u al applica ions, such as c op moni o ing and managemen o help ood secu i y
•De ailed ege a ion and o es moni o ing
•Obse a ion o coas al zones (ma ine en i onmen al moni o ing, coas al zone mapping)
•Inland wa e moni o ing
•Snow co e moni o ing
•
Flood mapping and managemen ( isk analysis, loss assessmen , and disas e managemen
du ing loods)
•Logis ic se ices o municipali ies and public au ho i ies
•T a ic secu i y ( oad condi ions, isibili y o a ic signs).
Au ho Con ibu ions:
Concep ualiza ion, C.M.T. and A.N.A.; Da a cu a ion, C.M.T.; Fo mal analysis, C.M.T.,
M.L., M.A. and K.B.; Funding acquisi ion, A.N.A.; In es iga ion, C.M.T. and A.N.A.; Me hodology, C.M.T., M.P.,
M.L., M.A. and K.B.; P ojec adminis a ion, A.N.A.; Resou ces, C.M.T.; So wa e, C.M.T.; Supe ision, A.N.A.;
Valida ion, C.M.T., M.P., M.L., M.A. and K.B.; Visualiza ion, C.M.T., M.L. and M.A.; W i ing—o iginal d a ,
C.M.T.; W i ing— e iew & edi ing, C.M.T., M.P., M.L., M.A., K.B., T.M. and A.N.A.
Funding:
The wo k was unded by Eu opean Commission h ough EU Li e+ p ojec MONIMET P ojec
(LIFE12ENV/FI/000409) du ing 2013–2017.
Con lic s o In e es : The au ho s decla e no con lic o in e es .
Re e ences
1.
Winga e, L.; Ogée, J.; C emonese, E.; Filippa, G.; Mizunuma, T.; Miglia acca, M.; Moisy, C.; Wilkinson, M.;
Mou eaux, C.; Wohl ah , G.; e al. In e p e ing canopy de elopmen and physiology using a Eu opean
phenology came a ne wo k a lux si es. Biogeosciences 2015,12, 5995–6015. [C ossRe ]
2.
Ah ends, H.E.; B ügge , R.; S öckli, R.; Schenk, J.; Michna, P.; Jeanne e , F.; Wanne , H.; Eugs e , W.
Quan i a i e phenological obse a ions o a mixed beech o es in no he n Swi ze land wi h digi al
pho og aphy: Phenology by use o digi al pho og aphy. J. Geophys. Res. Biogeosci. 2008,113. [C ossRe ]
Da a 2018,3, 23 18 o 19
3.
Ah ends, H.; E zold, S.; Ku sch, W.; S oeckli, R.; B uegge , R.; Jeanne e , F.; Wanne , H.; Buchmann, N.;
Eugs e , W. T ee phenology and ca bon dioxide luxes: Use o digi al pho og aphy o p ocess-based
in e p e a ion a he ecosys em scale. Clim. Res. 2009,39, 261–274. [C ossRe ]
4.
Ba e , C.W.; Coops, N.C.; Wulde , M.A.; Hilke , T.; Nielsen, S.E.; McDe mid, G.; S enhouse, G.B. Using digi al
ime-lapse came as o moni o species-speci ic unde s o ey and o e s o ey phenology in suppo o wildli e
habi a assessmen . En i on. Moni . Assess. 2011,180, 1–13. [C ossRe ] [PubMed]
5.
Linkosalmi, M.; Au ela, M.; Tuo inen, J.-P.; Pel oniemi, M.; Tanis, C.M.; A slan, A.N.; Kola i, P.; Aal o, T.;
Rainne, J.; Lau ila, T. Digi al pho og aphy o assessing ege a ion phenology in wo con as ing no he n
ecosys ems. Geosci. Ins um. Me h. 2016, 1–25. [C ossRe ]
6.
Meye , G.E.; Ne o, J.C. Ve i ica ion o colo ege a ion indices o au oma ed c op imaging applica ions.
Compu . Elec on. Ag ic. 2008,63, 282–293. [C ossRe ]
7.
Mizunuma, T.; Koyanagi, T.; Mencuccini, M.; Nasaha a, K.N.; Winga e, L.; G ace, J. The compa ison o
se e al colou indices o he pho og aphic eco ding o canopy phenology o Fagus c ena a Blume in eas e n
Japan. Plan . Ecol. Di e s. 2011,4, 67–77. [C ossRe ]
8.
Mo ohka, T.; Nasaha a, K.N.; Oguma, H.; Tsuchida, S. Applicabili y o G een-Red Vege a ion Index o
Remo e Sensing o Vege a ion Phenology. Remo e Sens. 2010,2, 2369–2387. [C ossRe ]
9.
Nijland, W.; Coops, N.C.; Coogan, S.C.P.; Ba e , C.W.; Wulde , M.A.; Nielsen, S.E.; McDe mid, G.;
S enhouse, G.B. Vege a ion phenology can be cap u ed wi h digi al epea pho og aphy and linked o
a iabili y o oo nu i ion in Hedysa um alpinum.Appl. Veg. Sci. 2013,16, 317–324. [C ossRe ]
10.
Pel oniemi, M.; Au ela, M.; Bö che , K.; Kola i, P.; Loeh , J.; Hokkanen, T.; Ka hu, J.; Linkosalmi, M.;
Tanis, C.M.; Me sämäki, S.; e al. Ne wo ked web-came as moni o cong uen seasonal de elopmen o
bi ches wi h phenological ield obse a ions. Ag ic. Fo . Me eo ol. 2018,249, 335–347. [C ossRe ]
11.
Richa dson, A.D.; Jenkins, J.P.; B aswell, B.H.; Hollinge , D.Y.; Ollinge , S.V.; Smi h, M.-L. Use o digi al
webcam images o ack sp ing g een-up in a deciduous b oadlea o es . Oecologia
2007
,152, 323–334.
[C ossRe ] [PubMed]
12.
Sonnen ag, O.; Hu kens, K.; Teshe a-S e ne, C.; Young, A.M.; F iedl, M.; B aswell, B.H.; Milliman, T.;
O’Kee e, J.; Richa dson, A.D. Digi al epea pho og aphy o phenological esea ch in o es ecosys ems.
Ag ic. Fo . Me eo ol. 2012,152, 159–177. [C ossRe ]
13.
Zhao, J.; Zhang, Y.; Tan, Z.; Song, Q.; Liang, N.; Yu, L.; Zhao, J. Using digi al came as o compa a i e
phenological moni o ing in an e e g een b oad-lea ed o es and a seasonal ain o es . Ecol. In o m.
2012
,
10, 65–72. [C ossRe ]
14.
Miglia acca, M.; Gal agno, M.; C emonese, E.; Rossini, M.; Me oni, M.; Sonnen ag, O.; Coglia i, S.; Manca, G.;
Dio i, F.; Buse o, L.; e al. Using digi al epea pho og aphy and eddy co a iance da a o model g assland
phenology and pho osyn he ic CO2up ake. Ag ic. Fo . Me eo ol. 2011,151, 1325–1337. [C ossRe ]
15.
Lopes, A.P.; Nelson, B.W.; Wu, J.; de Alencas o G aça, P.M.; Ta a es, J.V.; P ohaska, N.; Ma ins, G.A.;
Saleska, S.R. Lea lush d i es d y season g een-up o he Cen al Amazon. Remo e Sens. En i on.
2016
,182,
90–98. [C ossRe ]
16.
Yang, X.; Tang, J.; Mus a d, J.F. Beyond lea colo : Compa ing came a-based phenological me ics wi h lea
biochemical, biophysical, and spec al p ope ies h oughou he g owing season o a empe a e deciduous
o es : Seasonali y o lea p ope ies. J. Geophys. Res. Biogeosci. 2014,119, 181–191. [C ossRe ]
17.
Keenan, T.F.; Da by, B.; Fel s, E.; Sonnen ag, O.; F iedl, M.A.; Hu kens, K.; O’Kee e, J.; Klos e man, S.;
Munge , J.W.; Toomey, M.; e al. T acking o es phenology and seasonal physiology using digi al epea
pho og aphy: A c i ical assessmen . Ecol. Appl. 2014,24, 1478–1489. [C ossRe ] [PubMed]
18.
A slan, A.; Tanis, C.; Me sämäki, S.; Au ela, M.; Bö che , K.; Linkosalmi, M.; Pel oniemi, M.
Au oma ed Webcam Moni o ing o F ac ional Snow Co e in No he n Bo eal Condi ions. Geosciences
2017,7, 55. [C ossRe ]
19.
Be na d, É.; F ied , J.M.; Tolle, F.; G iselin, M.; Ma in, G.; La ly, D.; Ma lin, C. Moni o ing seasonal
snow dynamics using g ound based high esolu ion pho og aphy (Aus e Lo énb een, S alba d, 79
◦
N).
ISPRS J. Pho og amm. 2013,75, 92–100. [C ossRe ]
20.
Ga elmann, J.; Pohl, S.; Weile , M. F om obse a ion o he quan i ica ion o snow p ocesses wi h a
ime-lapse came a ne wo k. Hyd ol. Ea h Sys . Sci. 2013,17, 1415–1429. [C ossRe ]
21.
Hä e , S.; Be nha d , M.; Co ipio, J.G.; Schulz, K. PRACTISE—Pho o Rec i ica ion and Classi icaTIon
So wa E (V.1.0). Geosci. Model De . 2013,6, 837–848. [C ossRe ]
Da a 2018,3, 23 19 o 19
22.
Sal a o i, R.; Plini, P.; Gius o, M.; Val , M.; Salzano, R.; Mon agnoli, M.; Cagna i, A.; C epaz, G.; Sigismondi, D.
Snow co e moni o ing wi h images om digi al came a sys ems. I al. J. Remo e Sens.
2011
, 137–145.
[C ossRe ]
23.
Richa dson, A.D.; Hu kens, K.; Milliman, T.; Aub ech , D.M.; Chen, M.; G ay, J.M.; Johns on, M.R.;
Keenan, T.F.; Klos e man, S.T.; Kosmala, M.; e al. PhenoCam Da ase 1.0: Vege a ion Phenology om Digi al
Came a Image y, 2000–2015; ORNL DAAC: Oak Ridge, TN, USA, 2017. [C ossRe ]
24. Aus alian Phenocam Ne wo k. A ailable online: h ps://phenocam.o g.au/ (accessed on 9 May 18).
25.
Pel oniemi, M.; Au ela, M.; Bö che , K.; Kola i, P.; Loeh , J.; Ka hu, J.; Linkosalmi, M.; Tanis, C.M.;
Tuo inen, J.-P.; A slan, A.N. Webcam ne wo k and image da abase o s udies o phenological changes o
ege a ion and snow co e in Finland, image ime se ies om 2014 o 2016. Ea h Sys . Sci. Da a
2018
,10,
173–184. [C ossRe ]
26.
Tsuchida, S.; Nishida, K.; Kawa o, W.; Oguma, H.; Iwasaki, A. Phenological eyes ne wo k o alida ion o
emo e sensing da a. J. Remo e Sens. Soc. Jpn. 2005,25, 282–288.
27.
Mo is, D.; Boyd, D.; C owe, J.; Johnson, C.; Smi h, K. Explo ing he Po en ial o Au oma ic Ex ac ion o
Vege a ion Phenological Me ics om T a ic Webcams. Remo e Sens. 2013,5, 2200–2218. [C ossRe ]
28.
Jacobs, N.; Bu gin, W.; F id ich, N.; Ab ams, A.; Miskell, K.; B aswell, B.H.; Richa dson, A.D.; Pless, R.
The global ne wo k o ou doo webcams: P ope ies and applica ions. In P oceedings o he 17 h ACM
SIGSPATIAL In e na ional Con e ence on Ad ances in Geog aphic In o ma ion Sys ems, Sea le, WA, USA,
4–6 No embe 2009; ACM P ess: New Yo k, NY, USA, 2009; p. 111. [C ossRe ]
29.
Filippa, G.; C emonese, E.; Miglia acca, M.; Gal agno, M.; Fo kel, M.; Winga e, L.; Tomelle i, E.;
Mo a di Cella
, U.; Richa dson, A.D. Phenopix: A R package o image-based ege a ion phenology.
Ag ic. Fo . Me eo ol. 2016,220, 141–150. [C ossRe ]
30.
PhenoCam So wa e Tools. A ailable online: h ps://phenocam.s .unh.edu/webcam/ ools/ (accessed on
11 June 2018).
31.
Tanis, C.M.; A slan, A.N. Finnish Me eo ological Ins i u e Image P ocessing Toolbox (Fmip o ) Use Manual;
Zenodo: Genè e, Swi ze land, 2017. [C ossRe ]
32.
Coelho, L.P. Maho as: Open sou ce so wa e o sc ip able compu e ision. J. Open. Res. So w.
2013
,1, e3.
[C ossRe ]
33.
MONIMET EU Li e+ (LIFE12 ENV/FI/000409). A ailable online: h ps://monime . mi. i/ (accessed on
9 May 2018).
34.
Pel oniemi, M.; Au ela, M.; Bö che , K.; Kola i, P.; Linkosalmi, M.; Loeh , J.; Tanis, C.M.; A slan, A.N.
Da ashee o Ecosys em Came as Ins alled in Finland in Monime Li e+ P ojec ; Zenodo: Genè e, Swi ze land, 2017.
[C ossRe ]
35.
SPICE—Solid P ecipi a ion In e compa ison Expe imen . A ailable online: h ps:// al.uca .edu/p ojec s/
SPICE/ (accessed on 9 May 2018).
©
2018 by he au ho s. Licensee MDPI, Basel, Swi ze land. This a icle is an open access
a icle dis ibu ed unde he e ms and condi ions o he C ea i e Commons A ibu ion
(CC BY) license (h p://c ea i ecommons.o g/licenses/by/4.0/).