Detection of relative chlorophyll content of field cantaloupe canopy at different growth stages based on digital images
To explore the feasibility of quickly and non-destructively obtaining the relative chlorophyll content of the complete canopy of field cantaloupe,a self-made image acquisition vehicle equipped with a plane array camera was used in the present study to collect a total of 378 complete canopy images of field cantaloupe during the vine stretc-hing stage,the flowering stage,and the fruit expansion stage,under different water and fertilizer treatments.After image processing,32 color features and 6 texture features were extracted,and the correlation between image features and relative chlorophyll content of canopy was analyzed.After image features being preprocessed,principal compo-nents were selected as model inputs to establish multiple linear regression(MLR),support vector regression(SVR),and random forest(RF)prediction models for the relative chlorophyll content of field cantaloupe canopy at different growth stages.By comparison,it was found that the SVR models had the best performance,as the determi-nation coefficients of the regression within the predicted values and the measured values were 0.73,0.73 and 0.83,respectively,and the root mean squre errors were 0.90,0.91 and 0.76 for the vine stretching stage,the flowering stage and the fruit expansion stage,respectively.It was proved that digital image technology may enable rapid and non-destructive detection of relative chlorophyll content of the field cantaloupe canopy at different growth stages,which could provide technical references for the field management of cantaloupe production.
canopy image of cantaloupechlorophyllimage featuressupport vector regressioncritical growth pe-riod
万方数据
维普
