Classification of Varietal Type of Philippine Rice Grains Using Image Processing through Multi-View 3d Reconstruction

Abstract

Number of previous works have been carried out to study the classification of rice in the flat 2D image. In computer vision, 3D reconstruction is a field that has great theoretical research value. This paper attempts to see the performance by using features extracted from the 3D reconstructed image of a rice grain in classifying different varieties of it. We identify 250 rice grains belonging to five (5) varietal groups of rice in the Philippines. The data gathered was in the form of a video from a camera then it was split into images. After applying image enhancement operations, the images where used to build the 3D model using Agisoft Photoscan software. After the reconstruction, 18 color features and 21 texture features were extracted from the image. After the feature selection, 32 features were selected for this study. The proposed method uses BPNN architecture to classify the rice grain s varietal type.