Image detection and classification of particle-shape mixed agricultural products

There are lots of kinds of particle-shape agricultural products. It happens frequently when many kinds of products are mixed together. It is an important research topics that how to detect and classify every product. The paper realized image detection and classification of rice, soybean and mungbean mixed particle-shape agricultural products by computer visual technology. Firstly, original image is binarized. Adhering particles are segregated with morphological watershed algorithm. Edge of object connected region is extracted and labeled with eight neighborhood algorithm. Finally, shape and color features of every particle are extracted. Comparing extracted parameters and standard parameters of rice and soybean and mungbean has realized image detection and classification of mixed particle-shape agricultural products. The classification accuracy of the algorithm is 95.6%, and it offers certain guidance to automatic detection application of particle-shaped agricultural products.