Research on Tracking Recognition Algorithm of Steel Pipe Welding Seam Image

Against the inaccuracy and poor real-time of traditional image-identifying arithmetic of steel pipe welding seam, this paper offer a new method of disperse convolution. Firstly, a self-adaptive grey transferring method is put forward according to the disadvantage of traditional grey transformation method of fixed threshold, before filter function is established by the statistical characteristics of welding seam images and elements, and image-identifying arithmetic of disperse convolution is adopted to identify the welding seam center. Finally, the method, which has been widely used in industrial production, is validated to be of high anti-interference capability, high calculating speed, accurate allocation and easy implementation.