Abstract: In view of problems of low identification accuracy and efficiency existed in most recognition method of coal and gangue image which only used gray mean and gray variance, an extraction method of gray information and texture features of coal and gangue image was proposed. The representative feature parameters such as gray average, smoothness and energy, contrast, correlation, entropy of gray co-occurrence matrix are extracted, which are taken as the important basis for identification of coal and gangue. Matlab simulation results show that the gray information and the texture feature can well describe characteristics of the coal and the gangue image, which can provide an important reference for automatic identification and separation of coal and gangue.