Detecting Skin Defect of Fruits Using Optimal Gabor Wavelet Filter

In this paper we present a robust algorithm for automatic fruit defect detection. The proposed method works based on analysis of fruit skin as a texture image. First a bank of Gabor filters is applied on the fruit image. Then based on the response of filters the optimal filter is selected among the filter bank. By thresholding the response of the optimal filter, the skin defects are detected. The experimental results on different type of fruits show that the developed algorithm is robust, scalable and computationally efficient for detection of fruit defects. The accuracy of the method makes it also suitable for real-time sorting industry.