
Classification of agricultural products is a necessity for agricultural marketing to increase the speed and minimize the misclassification. One of the main problems in the automation of modern farming and production regards the selection of good fruits from the whole crop. The selection process in the fruits is very difficult in the market through manually. Some attempts at automatic classification using traditional computing resources and algorithms have been made. Unfortunately the average classification times for each orange using such methods are too long for an efficient real-time application. It is easily understood that the classification of fruits based on their digitized images can be improved and simplified if useless details are removed from these images. This paper is based on the phytochemical characterization of the bioactive compounds such that the reference image can be obtained more accurately, because the color and texture of fruits are the result of its composition. And also involves the development of algorithms for quality inspection of oranges by defining three quality classes (extra, I and II) for the fresh oranges. The fruit in the extra class must be of superior quality with no defects and high volume, whereas the classes I and II are moderately and highly defective fruits with less volume respectively. The methodology involves a two step process. The first step is to segment the image into defect and non defect classes by utilizing histogram based thresholding. According to second step three different approaches have been proposed. In the first approach of the second step texture features are extracted from GLCM. Second approach combines color and texture by extracting features from color GLCM.