The Research And Implementation Of Algorithms Of License Plate Localization And Character Segmentation

Video surveillance is an important approach to deal with security measures to avoid public damage and unauthorized access. With the advancement in technology, more and more people become aware of its importance, surveillance cameras are more effective than ever before, and protection provided by many high resolution ratio cameras are set in countless applications such as banks, airports etc. In video surveillance, the main monitoring subjects are the people. So, the results are affected by the people's subject factor. How to quality the computer to process the video surveillance is our main work. This subject is quite meaningful, and is classified to Smart Surveillance.This paper mainly introduces a new method to detect the activities which happens in the airport, including "Running person","Pointing" and "Person in yellow vest". We formulate different strategy for different action detection system, since each action has separate features. The overall scheme is quite similar as in the HOG features and SVM classifier. To detect the Running action, we employ the size of the rectangle; to detect the pointing action, we use Corner Detection method to help us find the movement of fingers; to detect the persons who in yellow vest appear in the video, we calculate the u-component of the image chroma. Finally, we testify our scheme with some videos, and based on the experimental results, detail analysis are presented. Moreover, we also discussed limitation of our method, and give some recommendations to further improve it.