Fourth-order B-spline wavelet multiscale local modulus maxima for edge detection in identification of pipeline fault

摘要

In this paper, for the actual in-service pipeline detection, an edge detection method based on fourth-order B-splinewavelet multi-scale local modulus maxima is proposed. We extract defect edge using wavelet maximum algorithm, select fine length , invariant moment , gray scale energy and so on, several key characteristic parameters that aregood for defect identification, and use BP neural network of single-output form for pattern recognition. By this method we complete the classification of the pipe weld joints and corrosion defects, and quantitative identification of corrosion defects.