The Rail End Face Contour Extraction Based on Mathematical Morphology and Adaptive Threshold

To solve the problem of extracting rail end face contour in high quality, this paper proposes a rail end face contour extraction algorithm based on mathematical morphology and adaptive threshold. First, use median filter on the rail end face image captured by the industrial camera, and then design a "(sic)" shaped kernel for image denoising based on mathematical morphology; finally, in contour extraction section, an improved Otsu algorithm combined with a histogram of gradient magnitude automatically determine the threshold. By double threshold detection, extract a complete clear contour of the rail end face. Experimental results show that the algorithm can not only suppress the smaller noise, but also have a good inhibition on noisy blocks. It is a strong self-adaptive algorithm, can get a great rail end face contour.