Abstract: This paper reported the way of defect extracting by wavelet modulus maxima method for welding area of computed radiography(CR) image, for which the defect extraction of aluminum welding CR image was taken as an example. Firstly, we used wavelet modulus maxima method to obtain enhanced grey curve of Al welding CR image which could show a clear boundary and local position for weld defect. Sequentially, we classified defect in welding zone as crack defect or non-crack defect by peak width, and compared it with other defect extraction methods. The results indicated that wavelet modulus maxima algorithm was feasible and practical, which improved the recognition accuracy of welding zone and defects extraction in computed radiography image based on wavelet transform modulus maxima.