Automotive Engine Cylinder Head Crack Detection: Canny Edge Detection With Morphological Dilation

Current inspection and checking of an automotive engine cylinder head crack usually involve manual inspection and mechanics checking dry magnetic powder, magnifying glass, and others. This is often labor-intensive, tedious, and involves a high degree of variability among mechanics. Therefore, developing an automated crack detection system will improve the safety and efficiency of data collecting, reliability, consistency, and accuracy. This article is a crack detection algorithm based on canny edge detection with morphological dilation techniques for automotive cylinders. The crack detection processing involves an image of a cylinder head taken and enhanced to 256 by 256 image size. The image is then transformed into a black and white image, and a Gaussian filter smoothes the cracked black and white image. We calculate an intensity gradient and magnitude using a filter to sharpen the edges and utilize a double threshold. Finally, we utilize a morphological dilation technique to extract the cracks from the entire engine cylinder head. The experimental outcomes show that our algorithms can extract cracks with accuracy. Moreover, the cracks can be effectively extracted from the image using our proposed techniques.