Accurate Identification Strategy of Coal and Gangue Using Infrared Imaging Technology Combined With Convolutional Neural Network

To effectively separate coal and gangue, accurate classification is an important prerequisite. Here, a new recognition solution for gangue proposed, in which the convolutional neural network (CNN) trained to achieve automatically identifying based on infrared images without considering selection of feature extraction classifier. Firstly, specific architecture detailed parameters model are optimized CNN only one Inception Block contains three different convolution kernels considered be most appropriate model. Next, performance proposed identification analyzed evaluated by image dataset, we discovered that capable correctly 192 training samples 48 test samples. Finally, compared with traditional other model, it proved has superior performance. The results state clearly combination can quickly accurately identify complex processing steps. At same time, certain anti-interference ability noises. And reference value research development intelligent preparation equipment.