Research on Recognition Algorithm Based on Polarization Characteristics of Coal and Gangue Surface

In this paper, an intelligent identification method of gangue based on polarization imaging technology is proposed. Firstly, a gangue image acquisition system was built in the laboratory, and 2000 groups of coal and gangue samples were collected, including polarization intensity images, polarization angle images and polarization degree images of coal and gangue; secondly, after preprocessing the polarization images to obtain the regions of interest of coal and gangue in the images, the grayscale characteristics of the polarized images, the grayscale features of the grayscale histogram features and the texture features of the polarized images were extracted respectively; finally, The Relief-SVM and the single hidden layer neural networks were used respectively to train the classification model. Experimental results show that the recognition rate of the images acquired by polarization cameras can reach 99.28% and 100% respectively. The accuracy rate is improved compared to the method of classifying ordinary black and white images of coal and gangue with non-polarization characteristics.