The Research And Realization Of Mineral Transport Vehicle License Plate Recognition System

Our country is the third largest mineral resources country in the world, which has full range of mineral resources and abundant storage, but its per-capita average of mineral resources amounts to only 58% of the respective world averages because of its huge population [1]. At present, the information level in the mineral management process is very low; most of the works in this process are completed by the staff's manual operation. It caused the statistical data distortion, tax loss, excessive exploitation of mineral resources, even damage to the natural environment. The mineral resources are a non-renewable resource, which play a vital role in the development of national economy. So, the information level in the mineral management process should be improved urgently in order to change the traditional production and the distribution management model, guarantee the mineral industry scientific and orderly development and make the mineral resources to be used rationally.Through by the actual investigation to the ticket-checking place at the mineral transport station which lies in the Renhe District, Yanbian County and Miyi County in the Panzhihua City, we found that there is a low recognition rate for the MTVLP (Mineral Transport Vehicle License Plate) and a lot of manual operation and it don't meet the needs of information, although the mineral transport vehicles are monitored. But the accurate and auto recognition for the MTVLP is the key point to improve the information level in the mineral industry.In this paper, we consider the particularity of recognition to the MTVLP picture, such as the different size and slanting angle of the MTVLP, the MTVLP has been polluted seriously and its picture's quality is poor and the weak work condition for the MTVLP Recognition System. And we study carefully and all-round on the each steps and difficulties with the license plate recognition. The main contents in this paper are as follows:(a)MTVLP Picture Location: On the basis of the study on the license plate location algorithms, we determined the method to the MTVLP picture's location and processed the rough positioning of the MTVLP picture by using a method proposed by this paper, which it is the accurate license plate location for mineral transport vehicle based on threshold segmentation. After the processing, some ineffective part such as the background noise and the plate border of the MTVLP are removed, only the effective parts—the license plate character of the MTVLP are reserved. This laid the foundation for preprocessing and recognition to the MTVLP.(b)MTVLP Preprocessing: The MTVLP pictures acquired in the actual environment are not affected by noises as a result of illumination and equipment, but it also polluted and destructed by a variety of factors which cannot be known in advance, and this reduced the quality of the picture. We did the study and experiment on the many different binaries'method and selected the image binaries'method which based on global mean and local standard deviation to do binary process for the MTVLP. Moreover, we proposed a noise removal method based on the connecting area labeling. The experiment show that this method can remove noises perfectly, it provides a good input for the recognition of the MTVLP characters.(c) MTVLP Characters Recognition: Considered the differences on the structure of the characters between the Chinese characters and the letters, numbers on the plate, we designed the Chinese characters classifier and the letters and numbers classifier individually. Specifically, the Chinese characters recognition used the method based on a minimum distance between the templates matching; the letters and numbers are be recognize by the method proposed in this paper, which based on the whole structural features of characters, especially the convex and concave feature of the characters. This method doesn't need to do the preprocessing for the license plate character such as slanting correction and thinning, so it can avoid losing the information of the character. This method not only can to improve the recognition rate, but also decrease the calculation and is a effective algorithm.(d)MTVLP System Realization: On the basis of the above contents, considered the completeness and flexibility of the LPR system, we realized this MTVLP System in the form of OCX controls, so this MTVLP System can be used conveniently, easily and repeatedly in differences programming language platform directly.