Research And Development On The Video Monitoring Equipment For Counting The Number Of People In Library

With the development of video monitoring equipment and the progress of image processing technology, video monitoring has become more intelligent and has been widely used in public management, security and precaution. As an important public place in university, how to take efficient management in library and how to make full use of library have always been an important aspect of research. The use of the library of existing monitoring equipment, using video image processing technology solutions, low cost, easy to operate, real good advantage. This paper is the use of the library surveillance video, the statistics the number of regions to make full use of library resources is of great practical significance.This paper analyzes the differences between libraries and other places, and compares advantages and disadvantages of several common solutions for counting the number of people in library. For library-specific needs, design a number of statistical systems surveillance video library, and ask questions encountered in the design of the system, gives ideas to solve these problems. Next comes the body features face detection processing algorithms to extract video human characteristics and uses AdaBoost based face detection algorithm statistical classification of human characteristics. Finally library unique environment, improve the human body feature detection algorithm, and the number of statistics in library systems.As libraries have special circumstances, first of all we have a large number of shelves in the library block of the control objectives and control objectives bookshelf next to the stand, followed by the library readers mainly sitting, but sitting direction and not uniform, and the object are different attitude, and finally the number density of each region within the library, light scenes big difference. For these situations, we propose the following appropriate solutions.I. According to the library scene monitoring equipment and the relative position of each body is different, relatively clear set of human characteristics to extract statistical number, effectively reducing the impact on the result of the detection due to the different environment brings.II. The same scene cascade approach, using different feature extraction methods and classification methods combined features while extracting more effective set of human characteristics, enhance the accuracy of test results.III. Due to environmental changes or disturbances, the initial data detected in the actual value will vary up and down. Based on the initial data, the average processing, effectively reducing the impact of test data drift, the end result is more stable and accurate..Experimental results of surface, this paper presents the number of library statistics for internal solutions to improve statistical accuracy. Finally, the paper shortage during the study were analyzed the presence and noted that the next work plan.