ElMasry, G., ElSheikh, I., Morsy, N. (2009). COLOUR GRADING OF STRAWBERRY USING COMPUTER VISION AND BACKPROPAGATION ARTIFICIAL NEURAL NETWORK. Journal of Soil Sciences and Agricultural Engineering, 34(6), 7063-7077. doi: 10.21608/jssae.2009.100732
G. M. ElMasry; I. H. ElSheikh; Noha E. Morsy. "COLOUR GRADING OF STRAWBERRY USING COMPUTER VISION AND BACKPROPAGATION ARTIFICIAL NEURAL NETWORK". Journal of Soil Sciences and Agricultural Engineering, 34, 6, 2009, 7063-7077. doi: 10.21608/jssae.2009.100732
ElMasry, G., ElSheikh, I., Morsy, N. (2009). 'COLOUR GRADING OF STRAWBERRY USING COMPUTER VISION AND BACKPROPAGATION ARTIFICIAL NEURAL NETWORK', Journal of Soil Sciences and Agricultural Engineering, 34(6), pp. 7063-7077. doi: 10.21608/jssae.2009.100732
ElMasry, G., ElSheikh, I., Morsy, N. COLOUR GRADING OF STRAWBERRY USING COMPUTER VISION AND BACKPROPAGATION ARTIFICIAL NEURAL NETWORK. Journal of Soil Sciences and Agricultural Engineering, 2009; 34(6): 7063-7077. doi: 10.21608/jssae.2009.100732
COLOUR GRADING OF STRAWBERRY USING COMPUTER VISION AND BACKPROPAGATION ARTIFICIAL NEURAL NETWORK
1Agricultural Engineering Department, Faculty of Agriculture, Suez Canal University.
2Food Technology Department, Faculty of Agriculture., Suez Canal University.
Abstract
Colour is often used as an indication of quality, ripeness and freshness for agricultural products including strawberry fruits. A laboratory computer vision system was established for colour grading of strawberry (Fragaria × ananassa) based on its ripeness stage. Colour features extracted from an image contained the brightness values of each pixel in the image; therefore these features represent the appearance of the fruits and strongly reflect ripening stage and firmness of the fruits. Colour of each fruit in the image was expressed using the average value of three channels (red ‘R’, green ‘G’ and blue ‘B’) of all pixels representing fruit in the image. In addition, to obviate illumination differences and to facilitate differentiation between tested fruits, the RGB components were also transformed to normalized RGB (r, g and b) and to CIE L*a*b* colour space. The most significant colour features were selected based on the analysis of variance (ANOVA) tests experienced on all samples. A backpropagation artificial neural network (BPANN) model was applied as a pattern recognition tool for classification purposes and for fruit firmness prediction using only the selected significant colour features. The efficiency of BPANN model in classifying fruits to six ripeness stages was 92.88%. Furthermore, firmness of strawberry fruits was predicted with correlation coefficients of 0.91 % and 0.89 % for training and validation sets, respectively.
View on SCiNiTO
Top