Machine vision based real time cashew grading and sorting system using SVM and back propagation neural network

Abstract

In today's world consumers are greatly aware about quality of food products. So there is a great need to build automated quality management systems. Benefits of automating the quality management include reduced production cost and overall improvement in quality. Nowadays great deal of research is going on in the area of machine vision based grading of food products. Grading and sorting of cashew kernels are done manually in most of the countries which is time consuming and expensive. In this paper a real time classification system to automatically grade cashew kernels based on their color, texture, size and shape feature are presented. Multisresolutional wavelet and Contourlet transform are used for extracting texture features. The images of the cashew kernel are acquired using a Charge Coupled Device (CCD) camera, and then the images are preprocessed using an efficient background subtraction technique. Then various external features are extracted using machine learning techniques. For the experimental study, cashew kernels of five different varieties are collected. SVM and back propagation neural network classifiers are used and their performance in terms of accuracy is observed.