A comparative analysis of color features for classification of bulk chilli

Abstract

The Paper work presents an approaches to classify chilli class from their bulk sample chilli images using RGB and HSI and L∗a∗b Model colour features. A rule based algorithm is implemented taking into account, best RGB, HSI and L∗a∗b colour features, 9 colour features were computed for R-(red), G-(green), B-(Blue), H-(hue), S-(saturation), I-(intensity), L-(brightness), a-(chromaticity layer red&green), b-(chromaticity layer blue&yellow) images from each image samples. Best features were used as an input to classifier and tests were performed to identify best classification model. R-Average, Hue-Average, a-average Hue-mean, L∗_mean, a∗_mean and standard deviation values are considered for Rule Based Classification, We have considered four different varieties of Chilli, with stalk and without stalk. The recognition rate for RGB colour features chilli with stalk is 70.% and for chilli without stalk is 85% is obtained. The recognition rate for HSI colour features, chilli with stalk is 80% and for chilli without stalk is 90% is obtained. The recognition rate for L∗a∗b colour features chilli with stalk is 85% and for chilli without stalk is 95% is obtained.