Visual Positioning and Recognition of Gangues Based on Scratch Feature Detection

Abstract

Received: 18 January 2019 Accepted: 20 March 2019 The coals and gangues in a raw coal image have similar visual features, due to the presence of coal ash on the surface. Thus, it is difficult to locate and recognize coals and gangues on the transmission line through visual recognition. To solve the problem, this paper proposes a visual positioning and recognition method for gangues based on scratch feature detection. Firstly, an image acquisition system was designed to capture the clear and suitable images. Next, scratched features were prepared on gangue surface with mechanical tools, laying the basis for visual positioning and recognition. Afterwards, the texture feature recognition method based on grey-level co-occurrence matrix (GLCM) was adopted to identify coal and scratched gangue blocks. The test results show that the GLCM correlation feature parameter is effective for scratch recognition. The parameter and the said method were proved effective through experiments.