Abstract: For problems of existing coal flow detection methods of belt conveyor such as susceptibility of detection accuracy to environment, complex realization process, long time-consumption of information extraction and so on, an adaptive coal flow detection method of belt conveyor based on machine vision was proposed. Firstly, the original coal transportation image of belt conveyor is enhanced by a fusion algorithm based on wavelet transform and segmented by OTSU algorithm into belt image and coal image. Secondly, the segmented coal image is processed by cavity filling, contour detection and area calculation to obtain area information of the coal image. Finally, a coal flow detection algorithm based on mathematical modeling is used to obtain coal flow detection value through calculating transient volume of coal. The test results show that the average detection time of the method is about 30 ms, and error between detection results and the measurement ones of electronic belt scale is about 5%, which meets real-time and accuracy requirements for coal flow detection of automatic speed control system of belt conveyor.