Research And Achievement Of The Diagnoses Of The Surveillance Video Image Quality

Monitoring system has become more and more important in nowadays security system. With the development of intelligent monitoring, the coverage of monitoring system as well as the number of cameras in the system have been increased significantly. It is more and more difficult to maintain the monitoring system. While image quality is an important requirement for the intelligent monitoring system, if it could not be solved in time when the unusual problem occurs, it may lead to economic losses. Because of the large scale of monitoring system, it is time-consuming and expensive for manual detect. Besides, manual detect can’t make sure that the image quality is usual all the time. It is necessary to design a detection module for the monitoring system to diagnose its image quality by itself.In this paper, a surveillance video image quality diagnose method based on target detection is proposed to detect the unusual problems in the video images. The unusual targets include snowflakes noise, no-video signal, image stuck, image missing, color cast and image fuzzy. We have made a deep research in both color cast and image fuzzy. Both algorithms have been compared with other methods and achieved good result. In the end of this paper, we made all the detective algorithms into reality and made up them in a detective module which can improve the detection efficiency and reduce the cost of manual inspection.