Development of an efficient approach for detection and measurement of crack length in ceramic tile manufacturing using millimeter-wave imaging. (July 2022)

• Record Type: Journal Article
• Authors: Bivalkar, Mandar; Agarwal, Smriti; Singh, Dharmendra
• Is Part Of: NDT & E international. Volume 129(2022)
• Publication Date: 2022-07
• Subjects: Millimeter-wave imaging -- Crack detection -- Frangi filtering -- Connectivity enforcement; Nondestructive testing -- Periodicals; Contrôle non destructif -- Périodiques; Electronic journals; 620.1127
• DOI: 10.1016/j.ndteint.2022.102656
• Languages: English
• ISSNs: 0963-8695

Abstract

Abstract: The detection of crack and their measurement in the packed ceramic tiles is an important procedure for quality monitoring and testing in the tile manufacturing industry. In the investigation of undercover targets high-definition (HD) cameras or surface crack detection methods like eddy current testing, etc. fail but millimeter-wave imaging can be useful. In our experimental work, cracked tiles are covered with the packaging cardboard sheet, and the unknown crack is detected using Millimeter-Wave Imaging. In this paper, an adaptive and statistical approach is proposed for crack detection and a computer-aided non-invasive technique is developed to measure a crack length. After critically analyzing various available signal and image processing techniques, a line enhancement known as Frangi filtering is selected for crack detection. In the current scenario; the decision parameter for the Frangi filter known as the Frangi scale ratio is required to be fixed manually, to minimize the selection error automation of this process is done using a Genetic Algorithm (GA). In literature, the focus is limited to only the detection of cracks in the tile, but the estimation of the crack length is also essential. The possible use of image morphing techniques for the measurements of the crack length is explored. It is observed that the conventional Hough transform fails to measure the length of the individual crack segment, hence the pixel connectivity enforcement for hole filling and local intensity distribution equalization is proposed to extract the approximate coefficients belonging to the crack segment. The raw data collected previously from our laboratory is used and processed using the proposed methodology, the obtained results are satisfactory and very encouraging as compared to earlier results reported in the literature.