APPLICATION OF THERMAL IMAGING OF WHEAT CROP TO ESTIMATE CANOPY COVERAGE UNDER DIFFERENT MOISTURE STRESS CONDITIONS

摘要

Thermal imaging cameras determine the temperature of the object by non-contact measurements and give temperature reading for each pixel of the image. This proximal remote sensing technique work with the same principle of spot pyrometers. Thus the thermal image directly gives the temperature of the crop canopy and provides a better distinctibility between the two classes i.e., leaf and soil using image classification techniques. In this study thermal imaging was used to determine the canopy coverage using image classification analysis. As a further application of this technology, an attempt was made to estimate the canopy coverage of the wheat crop grown under different moisture stress conditions. Thermal Images were analyzed with five different supervised image classification techniques namely Maximum likelihood, Mahalanobis, Minimum distance to mean, Parallelepiped and Support Vector Machine methods using ENVI - image analysis software. Results showed that the best estimation of canopy coverage was possible using Support Vector Machine method, due to its higher overall classification accuracy and Kappa coefficient. This is further supported by the statistical analysis based on the comparison with instrument (plant canopy analyser) observed LAI and digital image derived canopy coverage In general Support Vector Machine method estimated the wheat crop canopy coverage from the thermal image meaningfully with high R~2 value of 0.915 and with low values of RMSE and MBE. Thus the present study clearly showed that thermal image analysis could be applied as a non-destructive, rapid, proximal remote sensing technique to characterize the crop canopy temperature and estimate the canopy coverage of the wheat crop grown under moisture stress conditions.