Weld defect extraction based on adaptive morphology filtering and edge detection by wavelet analysis

One of the most key steps in X-ray automatic inspection and intelligent recognition systems is how to extract defects and detect their edges effectively. In this paper, a novel method of defect extraction based on the adaptive morphology filtering (DEAMF) is proposed, whose structuring elements can be changed with the sizes of defects adaptively. By this method, defects in X-ray weld inspection images are extracted with well-kept shapes and high speeds. Then according to the theory of edge detection based on wavelet transform modulus maxima, a locally supported wavelet with good antisymmetry is developed to extract edges of defects and the results are satisfying.