Machine Vision-Based Precision Detection of Circular Holes Using Canny Threshold Optimization and Zernike Moments

This study proposes a precision detection method that integrates Canny operator threshold optimization with Zernike moments to address the issue of low measurement accuracy associated with the manual inspection of circular holes in sheet metal during industrial testing. A complete automated measurement system was developed based on the MATLAB platform. First, adaptive median filtering is employed for image preprocessing, with superior performance in noise suppression and detail preservation validated through Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity (SSIM) metrics. Subsequently, Otsu's thresholding method achieves robust segmentation between target and background, laying the foundation for subsequent edge detection. An innovative adaptive threshold selection strategy for the Canny operator based on composite weight scoring was proposed during edge detection, significantly enhancing circular hole edges' continuity and geometric integrity. Finally, by integrating Zernike moments with sub-pixel localization technology, ultra-precise localization of edge points at the sub-pixel level was achieved. Experimental results demonstrate that the system achieves a measurement repeatability standard deviation of less than 0.02 mm and controls the absolute error within ±0.05 mm.This performance surpasses the ±0.3 mm precision requirement in industrial settings, providing an effective solution for automated quality inspection of sheet metal hole manufacturing.