144. Coal−Rock Interface Detection Using Digital Image Analysis Technique

Aiming at the defects in traditional coal−rock interface detection (CID) system, a new CID model was developed using the digital image analysis technique. First, image features of coal and rock were analyzed in detail and feasibility study of the detection method based on vision was conducted. Second, some rock and coal samples were collected from typical mines in China, and image acquisition was carried out in the laboratory within a simulated environment. Third, the color image of the sample was enhanced by a method based on RGB color-components, in which a new image was synthesized from the color-components with weighted sum method and the weighted coefficients were generated using back-propagation (BP) neural network. Finally, the synthesized gray image was segmented by using a rapid region-based segmentation algorithm and a ratio presenting content of the rock in the coal seam was calculated as the output of the CID system. Study results indicate a very low detection error. The proposed algorithm in this paper is valid for all types of coal mines. The model is required to be calibrated with respective image data taken from the working face.