Identification and feature extraction of drainage pipeline cracks based on SVD and edge recognition method

Abstract

In order to solve the limitations and accuracy of current closed circuit television system (CCTV) for underground drainage pipeline robots, this paper uses singular value decomposition (SVD) method to preprocess images, which can binarize images to separate cracks features. On this basis, the sliding window segmentation and edge recognition method are used to realize the automatic identification and feature extraction of the cracks in the drainage pipe. The experimental results show that the proposed crack defect identification and feature extraction method can effectively identify the pipeline crack defects and feature extraction, and this method improve the efficiency and reliability of CCTV video detection in drainage pipelines.