Anisotropic diffusion based denoising on concrete images and surface crack segmentation

Abstract

Purpose Health monitoring of concrete is one of the important tasks in the structural health monitoring. The life of any infrastructure relies on the quality of the concrete. The computer vision-based methods are very useful to identify the structural defects. The identification of minor cracks in the noisy concrete image is complex. The purpose of this paper is to denoise the concrete crack images and also segment the cracks. Design/methodology/approach The novelty of the proposed work lies on the usage of anisotropic diffusion filter in the noisy concrete images. Initially anisotropic diffusion filter is applied to smoothen the concrete images. Adaptive threshold and gray level-based edge stopping constant are used in the diffusion process. The statistical six sigma-based method is utilized to segment the cracks from smoothened concrete images. Findings The proposed method is compared with five state-of-the-art-methods with the performance metrics of mean square error, peak signal to noise ratio and mean structural similarity. The experimental results highlight the advantages of the proposed method. Originality/value The novelty of the proposed work lies on the usage of anisotropic diffusion filter in the noisy concrete images. This research work gives the scope for structural damage evaluation by the automation techniques.