Application of chlorophyll fluorescence imaging technique in analysis and detection of chilling injury of tomato seedlings
Zhenfen Dong,Yuheng Men,Zhenzhen Liu,Jinpeng Li,Jianwei Ji
DOI: https://doi.org/10.1016/j.compag.2019.105109
IF: 8.3
2020-01-01
Computers and Electronics in Agriculture
Abstract:Low temperature is one of the main environmental factors limiting plant geographical distribution and crop production. Chilling injury is a stress caused by low, nonfreezing temperatures (0–12 °C). Since chilling injury is an important facility agriculture problem in northeast China regions, we aimed to study a suitable method for evaluate the potential of chlorophyll fluorescence imaging for the detection of chilling injury in tomato Seedlings. Here, four types of chlorophyll fluorescence image features were used to study and analyze chilling injury classes: chlorophyll fluorescence parameter values, histograms, textures and color descriptions. The Pearson correlations of the features in each type were calculated, and eligible features were input into BPNN model to identify chilling injury classes of leaf. Our results reveal six fluorescence parameter values of Y (II), qP, qL, Y (NPQ), Y (NO) and Fv/Fm can be used to evaluate the chilling injury classes of tomato seedlings. The identification accuracy rate of training set was 90.3%, and that of the validation set was 90%. The Skewness of the gray histogram features and standard deviation of entropy of the texture can be used to evaluate the chilling injury classes of tomato seedlings. The identification accuracy rates of training sets were 88.1% and 72.6%, respectively, and that of the validation set were 82.5% and 61.7%, respectively. The B, b and L/b of the color descriptor features can also be used to evaluate the chilling injury classes of tomato seedlings. The identification accuracy rates of the training sets were 89.8%, 91.9% and 90.1%, respectively, and that of the validation sets were 87.5%, 90.8% and 90.0%, respectively. Our research showed that the best recognition accuracy was b and L/b of the color descriptor features, as well as fluorescence parameter values, and the recognition accuracy are all more than 90%. Therefore, chlorophyll fluorescence imaging was recommended for the detection of chilling injury classes of tomato seedlings under low temperature stress, and had a good evaluation effect. The data reported in this manuscript should foster new research method for the detection of the chilling injury classes and have considerable prospects for non-destructive diagnosis of plant chilling injury.
agriculture, multidisciplinary,computer science, interdisciplinary applications