The Silicon Steel Based On Machine Vision Defect Detection Positioning Technology

The silicon steel as an important component of internal motors in industrial production, how to improve the detection efficiency of their products has become the Silicon steel defect inspection of the important content. Silicon steel in the production line, however, the time of placement, can not be placed in the right direction in full accordance with the transmission line, in order to follow-up to a higher level of treatment, it is necessary to locate the image into the black piece into the wrong location of the correct location. Position error performance for a very wide range of situations, mainly horizontal position is not correct, the direction is not correct. Image line detection algorithm can detect the straight line through the slope of the line defining the direction of the rotation angle, arc detection and identification algorithm can detect the arc of the center of a circle, will not correct the black horizontal position Katahira to the correct location.Silicon steel location for the defect detection techniques, this paper conducted a more in-depth research, mainly divided into the following sections:First,Straight-line location recognition algorithm research. Linear positioning the main study used a wider range of linear recognition algorithm for Hough transform, Hough transform pairs of random noise, and some cover the phenomenon is not sensitive, and has a strong anti-jamming capability, allowing detection of the boundary curve to be discontinuous.Second,The arc of the location recognition algorithm research. Change by studying classical Hough transform circle detection algorithm and random Hough transform algorithm to analyze the strengths and weaknesses of the two algorithms.Third,Silicon steel image to move to the standard position. By straight-line recognition algorithm to determine the location of the angle of rotation, and through the arc centered on the positioning of recognition algorithm to determine the amount of image shiftFourth,Algorithms for the inadequacies of arc made to improve algorithms. Hough transform algorithm for the above-mentioned shortcomings, propose an improved fast random Hough transform algorithm and improved four-point randomized Hough transform algorithm.Fifth,Silicon steels when the defective preliminary judgments recognition. Hough Transform line detection algorithm arc central angle, the initial judge was sufficiently flawed by improving fast randomized Hough transform algorithm and improved four-point randomized Hough transform algorithm to determine the results of the initial arc is flawed.