Based On Mathematical Morphology Image Segmentation And The Application In License Plate Segmentation

Character recognition of vehicles' license plates is the important application of computer vision and pattern recognition in the intelligent transportation system and the important step of intelligent transportation. The getting of vehicles' plates is the key of vehicles' plates recognition system. Because of the complex of image background, the uncertainty of plate position and unforesight of image quality, The character recognition of vehicles' plate is not satisfied. Using new theory and new technology to segment vehicles' plate is one of research keys. Mathematical morphology is the new science of researching image. It is building on assemble algebra' and using assemble theory ways to describe geometry structure. It researches image's geometry structure. The principle idea is using a structure element of carrying object character to collect image information. Because of the special advantage that structure element has, mathematical morphology image process becomes a important research field of computer digital image process.In this paper, The segment technology based on mathematical morphology is saved as the research field, used to solve vehicles' plate segmentation under error and improve the recognition ability of total system. The basic mathematical morphology operation and useful algorithm are researched and realized. After researching the kinds of filter algorithms and structure element characters, a kind of improve series structure element filter algorithm is proposed. It can't only clear dissociate error, but also restore the " sand holes " to improve the quality of image. Based on the compare of test result of edge detections in vehicles' plate image, vehicles' edge detection ways based on order morphology change is proposed and used to improve detect result that exit error. After kinds of research, the new way based on Top-hat license plate fast search and location is proposed. The way can fast locate every angle of license plate, examine first knowledge of license plate can also be used search process and the result of segmentation is the recognition. Because the way don't depend on any angle, and can search the region of the angle, it can solve the recognition of incline and lack angle license efficiently.In this paper, the way of segmentation and dictation has reference worth to the object that geometry shape has known.