Extraction of Weld Seam Centerline Based on Improved Canny Algorithm

Based on the analysis of characteristics in weld seam images, this paper proposes an improved method for detecting the centerline position using an adapted Canny algorithm, addressing issues such as susceptibility to noise interference and poor connection of strong and weak edges. Initially, adaptive median filtering and enhanced bilateral filtering replace traditional Gaussian filtering in the Canny algorithm. Subsequently, histogram analysis and Otsu's method are employed to adaptively select thresholds for the Canny operator. Finally, the Canny algorithm is utilized for extracting weld seam edges. Experimental results demonstrate that compared to Gaussian filtering, employing adaptive median and bilateral filtering improves Peak Signal-to-Noise Ratio (PSNR) by 9.2145 dB and reduces Mean Squared Error (MSE) by 33.32819, indicating superior effectiveness in preserving image details and eliminating noise. Pratt's Figure of Merit (PFOM) and Structural Similarity Index (SSIM) values approaching 1 further underscore the algorithm's enhanced precision.