Coal rock fracture detection is the key way to develop coalbed methane.In order to detect the fracture of the coal rock image,a method of combining multi-structure elements morphology with self-organizing map (SOM) neural network clustering is presented in this paper.Firstly,multi-structure elements morphological edge detection algorithm is used to detect all edges of the coal rock image,and then the feature parameters of coal rock image edges are calculated as the inputs of SOM neural network.Finally,the fracture edges and the non-fracture edges are classified by the SOM-based clustering algorithm,and the fracture edges of image are gained.Experimental results show this new method can effectively detect and extract the fracture information of coal rock image.