Published in Scientific Papers. Series B, Horticulture, Vol. LXVIII, Issue 1
Written by Mima TODOROVA, Stefka ATANASSOVA, Dimitar YORGOV, Petya VELEVA, Dimitar ZHELYAZKOV, Antonya PETROVA, Velika BOTEVA
The present study aimed to establish the relationship between the characteristics of RGB images of tomato fruits of three cultivars and important indicators of the quality of these fruits. In the study, 30 fruits from three cultivars at different ripening stages were used. The digital images of each tomato from four sides were obtained using a document camera. The same fruits were analyzed for dry matter, total dyes, vitamin C, titr. organic acids, beta carotene and lycopene content. The data of each image's G, R, and B channels were extracted and averaged. The average data of grayscale were taken as well. Based on the color characteristics and the chemical analyses, a procedure of descriptive statistics, correlation, and regression analysis was performed. The image procedure of all obtained data was performed using Jupyter notebook. The highest correlation coefficient (-0.860) was obtained between the Gray channel and total dye content, followed by the strong correlation between the Green channel and lycopene content (-0.820). And the models with the highest predictive ability were the models for total dye by Gray channel values (R2 = 0.815) and lycopene by Green channel (R2 = 0.761).
[Read full article] [Citation]
