Abstract: The problems with the coal flow segmentation methods currently available are inadequate acquisition of coal flow contour and edge information and poor in environmental adaptability. To address the problems, a coal flow segmentation method based on 4-frame difference and entropy fusion is proposed, which can detect whether a chute is blocked by coal based on coal flow profile area characteristics. The information of coal flow in state of motion is first acquired by using the frame-difference method, and then the local regional entropy value is calculated according to entropy of information. After the contour of coal flow is made known based on local entropy value image of coal flow which is obtained from the difference image, the detection result can be obtained through taking the foreground area as the feature and making classification of the feature using SVM. In order to evaluate the performance of the proposed method, test is made with the method under the circumstances with varying coal flow volumes and rates, natural light intensities and coal handling patterns. Result of analysis and comparison shows that with the use of the method, all the information related to the coal flow under different conditions can be accurately acquired. Moreover, the method is high in adaptability and capable of eliminating the interference of environmental illumination.