Research And Implementation Of License Plate Recognition System

The recognition of license plate has a very important application value in the intelligent transportation system.It is the foundation and key of the intelligent transportation system.Many experts and scholars at home and abroad have carried out a deep research on the recognition of license plate,and many theories and algorithms with outstanding contributions have emerged.License plate recognition combined with many technologies such as image processing and pattern recognition,can be widely used in parking lot management and high speed charge management,which saves a lot of manpower and material resources.This paper mainly studies the three essential links of license plate recognition system,including license plate location,license plate segmentation and license plate recognition,and implements a simple license plate recognition system by using C language and OpenCV library.Summary of the full text of the work,there are the following points:First,by describing the background and significance of the license plate recognition system,and by describing the research status at home and abroad,the research in this field has certain practical significance in China.Secondly,based on the analysis of the advantages and disadvantages of several commonly used algorithms,we get the license plate location algorithm based on the combination of color partition and projection technology.An experimental analysis of the location of the algorithm is carried out,and the positioning results are reliable.In the aspect of character segmentation,the character segmentation based on projection method based on connected domain segmentation,and analyzes the advantages and disadvantages,proposed segmentation method of vertical projection and prior knowledge based on the combination of characters,and to test the optimization steps and results,the character segmentation method of this system was reliable.Thirdly,the method of license plate character recognition is studied and designed.The advantages and disadvantages of segmentation algorithm based on connected domain,segmentation algorithm based on BP neural network and templatematching algorithm are analyzed.A recognition method based on improved multi template matching is proposed,and the recognition results are tested.It is feasible to determine the character recognition method of this system.Finally,it is designed and implemented for the system.In this chapter,a simple image based license plate location,segmentation and recognition system is implemented in this chapter.And the experiment of identifying the results of the system is carried out,and the experimental data are analyzed,and the recognition success rate of the system is obtained.By analyzing the data of license plate recognition under different conditions,an improved recognition system based on the precise location algorithm and the improved recognition algorithm is proposed.