Classification of Rice Quality Using Backpropagation Based on Shape and Color

The distribution of mixed rice on the market makes it difficult for consumers to determine the rice quality. In determining rice quality, the consumers consider and compare the texture, size and shape, color, aroma, purity, and homogeneity manually. This process is prone to errors and mistakes, due to the limited ability of each human's vision. Therefore, a method to determine the quality of rice automatically based on the physical characteristics of rice is needed. In this paper, we proposed an automatic rice quality classification method using backpropagation based on the shape and color of the rice. There are four parameters used to determine the classification process, namely compactness, circularity, mean, and skewness. Compactness and circularity were used to determine the ratio between the whole rice and broken rice. While mean and skewness were used to determine the color distribution of the rice. Experiments have been performed on 100 images consisting of 50 premium and 50 medium rice images. The experimental results show that the proposed method can classify rice based on its shape and color effectively with an accuracy rate of 95%.