Image de-noising on strip steel surface defect using improved compressive sensing algorithm

Abstract

De-noising for the strip steel surface defect image is conductive to the accurate detection of the strip steel surface defects. In order to filter the Gaussian noise and salt and pepper noise of strip steel surface defect images, an improved compressive sensing algorithm was applied to defect image de-noising in this paper. First, the improved Regularized Orthogonal Matching Pursuit algorithm was described. Then, three typical surface defects (scratch, scar, surface upwarping) images were selected as the experimental samples. Last, detailed experimental tests were carried out to the strip steel surface defect image de-noising. Through comparison and analysis of the test results, the Peak Signal to Noise Ratio value of the proposed algorithm is higher compared with other traditional de-noising algorithm, and the running time of the proposed algorithm is only26.6% of that of traditional Orthogonal Matching Pursuit algorithms. Therefore, it has better de-noising effect and can meet the requirements of real-time image processing.