Abstract: Coal gangue sorting is the most basic, effective, and important technical measure to improve coal quality. Improving the accuracy and efficiency of coal gangue sorting is a serious challenge faced by coal gangue sorting. In-depth research and analysis have been conducted on the existing intelligent sorting systems for coal gangue, including “grabbing sorting”, “fork sorting”, and “pneumatic sorting”. The academic thought of “Sorting gangue is sorting images” has been proposed, and the logical framework of the academic ideology of “Sorting gangue is sorting images” has been established. The basic connotation of the academic ideology of “Sorting gangue is sorting images” has been elucidated, mainly including image-based coal gangue recognition, image-based coal gangue sorting feature extraction, image-driven sorting machine dynamic target tracking, and multi-task multi-sorting machine collaboration based on image sequences. Aiming at the problem of image-based coal gangue recognition, a recognition principle and metsorting rates visual images and X-ray images is proposed, which can effectively improve the accuracy of coal gangue recognition; Aiming at the problem of feature extraction in coal gangue image sorting, a plane and depth feature extraction and fusion algorithm based on coal gangue images is proposed. A coal gangue sorting cube is constructed, which can improve the accuracy of coal gangue sorting; An image-based coal gangue matching tracking and path planning method is proposed for dynamic coal gangue tracking, which can improve the accuracy and reliability of sorting; Aiming at the problem of intelligent collaborative sorting of multiple sorters, it is proposed to construct a comprehensive benefit function of sorters based on the coal gangue image information database to achieve optimal allocation of multiple tasks for multiple sorters, and to integrate reinforcement learning methods to achieve intelligent collaborative control of multiple sorters and optimal configuration of the number of sorters. This can effectively improve the sorting efficiency of the multiple-sorter system. Following the academic ideology of “Sorting gangue is sorting images”, our team independently developed a double robotic arm truss type coal gangue sorting robot experimental platform, which verified the correctness and feasibility of this academic ideology and successfully applied it in Yuhua Coal Mine of Tongchuan Mining Company. The academic ideology that “Sorting the gangue is sorting the images” has laid a theoretical foundation for solving the problems of intelligent, precise, and efficient coal gangue sorting.