Abstract: In view of problems of no representation and practicality because that traditional separation system of lump coal and gangue based on image gray information only randomly selects samples of coal and gangue, the paper researched an automatic separation system of lump coal and gangue based on image gray information which emphasized in practicality and system. At first, it selected six representative lump coals and gangues in term of material, texture feature and cleanness, and analyzed distribution law of image gray of the six lump coals and gangues according to certain group mode under conditions of strong light, natural light and energy-saving lamp light and made the distribution law as basis of object recognition of computer, so as to realize automatic separation for lump coal and gangue. The test result showed that using obvious differences of image gray information between lump coal and gangue to make visual distinction can realize automatic separation under special conditions.