Abstract: The quantitative analysis of two-dimensional cracks in cataclastic coal is important for the effective prediction of the permeability and the strength of coal mass of coal resevoir in structurally complicated areas. Aiming at the bitmaps of two-dimensional micro cracks in cataclastical coal with different deformation, on the basis of MATLAB image processing platform, through binaryzation, noise reduction and restoration, programming realized automatic recognition of cracks, eliminated succesfully the disturbance of stria and mottle unequal in size, the recognized cracks tended to be accordant to the original cracks. Then on the basis of crack partitioning, the automatic extraction of parameters such as area, length, mean width and surfacial crack rate was realized. The crack information taken from two bitmaps was compared, the result was accordant to the visual development degree of cracks.