Abstract:

The rapid methods detecting chlorophyll concentration by the computer vision technology，and a unary quadratic model to predict chlorophyll content based on color parameters of tomato leaf images have been established in this study．The images of tomato leaves were taken in the image acquisition system，then the color characteristics were extracted with the MATLAB image processing software．The correlation between color parameters of tomato digital image and chlorophyll content of tomato functional leaf were analyzed by nonlinear regress models．The results showed that the color characteristics such as R/G、(G-R)/(G+R)、G-R、r、r-g in the RGB color system，and H-value in the HIS color system were significantly correlation with chlorophyll content of tomato leaf at . 6 sets of prediction model were established and among them 3 models with high fitting degree were selected to use．The prediction accuracy of the selected model were tested，and error ranged 0 to 22.22%．According to the determination coefficients and RMSE (root mean square error)，G-R was the best color characteristic to predict chlorophyll content of tomato leaf. The corresponding models are Chl a = 0.0926 + 0.1208 (G-R) - 0.0009 (G-R)2，Chl b = - 0.0252 + 0.0397 (G-R) - 0.0003 (G-R)2 and Chl a+b = 0.1271 + 0.1600 (G-R) - 0.0011 (G-R)2.

Key words: tomato, Machine vision, chlorophyll content, tomato leaf, color characteristic