Neural Network based on GLCM, and CIE L*a*b* Color Space to Classify Tomatoes Maturity

Tomato (Lycopersicum commune) is one of the fruits that can be consumed directly or processed into food or drinks. In the harvesting season, tomatoes will be picked and sorted into different class depending on its size, shape, and color. The quality and the taste of tomatoes fruit depends on the maturity level of this fruit. There is some product that used a different level of tomato maturity. Therefore it is necessary to classify the ripeness of this fruit. The use of information technology such as computer vision can help the farmer to classify tomato based on maturity level. To be able to determine the ripeness of tomatoes from an image, a computer needs to identify the right parameters to determine the maturity of this fruit. The texture and color on the surface of tomatoes is a feature that easily recognized in determining the maturity of tomatoes. Therefore this research proposed the combination of GLCM texture extraction and CIE L*a*b* color space and also Backpropagation neural network (BPNN) to classify tomatoes maturity. GLCM feature extraction consisted of Contrast, Correlation, Energi, Homogeneity, and Entropy. GLCM was implemented in each RGB color layer, then the average of CIE L*a*b* was calculated. From 75 train set and 25 test set, BPNN model can classified tomatoes maturity with accuracy 100% and average accuracy 95.46%.