Defect extraction techniques are studied regarding the characteristic of X-ray images of carbon product, and threshold-construction method based on iteration and edge-extraction algorithm based on mathematical morphology are advanced. In order to extract defects quickly and exactly, target boundary extraction algorithm and image enhancement algorithm based on wavelet transform are proposed, background removal and enhancement of object region are implemented successfully. Based on this method, combining mathematical morphology and iteration threshold segmentation is adopted to extract defect in order to eliminate the noise disturbance, and with iteration threshold segmentation, defect edge-extraction is realized based on edge-extraction algorithm of mathematical morphology. The experimental results indicate that the method can achieve automatic extraction of defect region and edge with weak noise disturbance , which lays a good foundation for flaw feature parameter extraction and choice.