Classification of rice seed MR219 and MR269 varieties based on morphological features using machine vision technique

Abstract

This project was carried out to classify the morphological features of rice seed varieties of MR219 and MR269 using machine vision technique. The study starts from acquiring image of the rice seed varieties of MR219 and MR269 using a CCD camera. The CCD camera was enclosed in a black box equipped with an illumination system. The images were processed and the morphological features were extracted from the image. This process was carried out in the LabVIEW software. The data was analyzed using MATLAB Neural Network. Through this thesis and data collected, the rice seed varieties can be determined and classified based on the extracted morphological features. The extracted of morphological features were the length, area, width, major axis length, minor axis length, thinness ratio, aspect ratio, rectangular aspect ratio, equivalent diameter, and extend. From the data collected, the lowest MSE which is 2.37592e-1 was acquired using 5 hidden layer of neuron with the highest classification accuracy which is 63.3%. However, the 5 hidden layer of neuron was then tested again to acquire a higher accuracy and the final MSE was 2.09587e-1 with an accuracy of 70%.