Coal gangue recognition is an important part of the coal intelligent industry, and the coal gangue recognition technology based on machine vision is the current research focus and hotspot, but there are some problems and deficiencies in the existing technology. Firstly, the influence of external factors on coal gangue image acquisition was revealed. Secondly, the common methods of coal gangue image processing were expounded and the shortcomings of each method was pointed out. Finally, the current recognition and classification technology of coal gangue based on machine learning was summarized. Based on the above research results, the further research directions of the current machine vision recognition methods for coal gangue are proposed, namely, building high-quality coal gangue datasets, researching efficient image segmentation methods, selecting the most representative coal gangue features, and optimizing the deep learning network model structure.