2025 年 12 巻 1 号 p. 24-00311
During the acquisition of worm gear profile images, the imagery is frequently compromised by Gaussian noise, variations in illumination intensity, and surface contaminants, which critically impair the subsequent extraction of worm gear profile details. To address these challenges, this study introduces an enhanced method for designing a worm gear tooth profile edge-preserving filter aimed at image detail enhancement. To amplify image details, a two-stage guided filtering algorithm is employed, utilizing the original image and the post-threshold segmented image as guide images for separate filtering processes, followed by a fusion step. The fused image undergoes additional edge enhancement through sharpening and contrast improvement via the adjustment of gray-level distribution. Experimental outcomes indicate that, compared to conventional edge-preserving filtering algorithms, the proposed method achieves a 22.45% increase in peak signal-to-noise ratio and a 20.1% improvement in image contrast. The denoised worm gear tooth profile edge image exhibits enhanced feature details while mitigating the impact of surface stains and uneven lighting on edge detection, resulting in clearer and more reliable detection outcomes.
