The Research Of Network Surveillance System Based On Fisheye Camera

In recent years, with the development of the communication technology, computer technology and multimedia codec technology, the video surveillance technology has been achieving rapid development and is widely used in the production and living of people, such as in the school, government, financial system,transportation, medical system, large factories and other fields. With the continuous expansion of its market application, new demand has been put forward, such as large-range without blind area real-time monitoring. For the shortage of traditional monitoring cameras’ small view, so in order to achieve real-time non-blind area,we need to rotate a single lens mechanically or stitch images captured by multiple cameras.These solutions lead to problems inevitably such as the increasement in the number of cameras and make it difficult to control the lenses’ exposure time.Fisheye lens has short focus and the wide-view approximate or more than 180 degree, which makes it become the best selection in building the large-view non-blind real-time surveillance system. In the moniter client,we select the fisheye lens as the camera.When the fisheye lens is installed on the roof, the camera can capture images during the 360 degree in horizontal and 180 degree in the vertical directions.We build the network video surveillance system,and in the client we realize to receive, decode, correct distortion and display the video data.This article has the following parts:(1) first introduce the research background, significance and research status;(2) put forward the overall framework and scheme of the network video monitoring;(3) analyze the principle of fisheye lens imaging, and simulate several fisheye distortion correction algorithms and analyze the results, then put forward the necessity to establish a new correction model;(4) a new correction algorithm based on this topic is put forward, including the extraction of feature points,getting rid of the false feature points, the establishment of the model, image interpolation, and the display model of corrected images;(5) discuss the network transmission technology, including the selection of network transmission protocol and the network architecture;(6) the design of client based on Directshow technology, including receiving and decoding of the video streaming data, the program design of panoramic view and small view of the corrected images;(7)at last,show the experimental results, summarize the work and put forward the further prospects.The experimental results show that the correction algorithm put forward in the article on the resolution of1600 * 1200 fisheye images works well.Also the processing time correcting single frame is no more than 35 ms which makes the video display smoothly and achieve the real-time performance. Moreover,we roam through the human-computer interaction,and the correction images can display in different pattern.The network surveillance system achieves the desired design effect.