Research On Edge Detection Technology Applying In Car Lock Identify And Print Code System

Edge detection is important in images processing, it is a problem which is not resolved primely to this day. It is by reason of the importance of the problem itself, on the other hand, it also reflects the difficulty and depth of this problem. Edge is basic feature of digital image, it saves the shape information of objects in image while it highly reduces the information need to be disposed. Edge detection includes measurement, detection, orientation of gray-change. Edge detection is playing a very important role in figure detection, image segmentation, pattern recognition and machine vision.Image edge detection has a long research history as a disposing process of low-level vision, new theories and methods emerge in endlessly, but there is no method can fit all images as yet because of the difference of the images feature.So we need suited algorithm for different application environment.This thesis sets about research for need of project, in the process of designing Car Lock Identify and Print Code System, knowing this system require to detect the figure of the keys quickly and accurately. For measuring parameter of key width, this system requires to get the single-point and continuous figure of the keys. Considering the dust on product line, algorithm should eliminate noise influence effectively. Based on above-mentioned character, this thesis intend to design an suited algorithm to resolve practical problems.So after introducing domestic and overseas research actuality about edge detection and designing programs for some representative arithmetic, comparing the result, this thesis brings forth an image figure tracking algorithm, and applies it in the project. The result indicates that the detected single-point edge has no fake edge, smooth and continuous, at the same time it has more abundant details of edges and has good localization. Especially, the algorithm can execute quickly, it can eliminate noise influence effectively. These superiorities establish good base for image scale measure. The efficient of this method is proved by practices. It is fit for the car lock identify and print code system. What's more, the algorithm can practice widely in scale measure of single-target image.