Design And Implementation Of The Real-Time Detection System For Surveillance Video Quality

In recent years, video surveillance technology has been widely used in intelligent transportation, security monitoring and other industries. With the drastic increase in the number of video surveillance equipment, it’s becoming more and more difficult to maintain and manage these equipment. Conventional management model with manual inspection surveillance is no longer applicable. Therefore an intelligent video surveillance system for automatic detection is of very important significance.This paper designs and implements a real-time video quality inspection system to solve the video quality analysis problems in video surveillance detection, the main work includes the following aspects:(1)Based on the actual needs of the video quality testing, the framework of the entire system is designed encapsulating image preprocessing, anomaly detection, results output and management. Wherein the image preprocessing module mainly deals with video format, resolution, etc.; anomaly detection module accomplishes detecting algorithm of common video anomalies; the result output module visualizes the results of detection module; management module is designed for users conveniently managing the system.(2)Analyse common abnormalities in video surveillance systems, such as signal missing, snow noise, sharpness, tint, and judge them in accordance with corresponding detection algorithm. Modified detection algorithms are proposed for image clarity and color cast that reflects image quality; According to the specific characters of clarity, great average gradient distinction and small calculation, an average gradient detecting algorithm based on modified focus window method is presented; respect for the cast:According to cast the image pixels adjacent video frames little change in the correlation characteristics, combined with the color change, this paper proposes a detection algorithm based on the covariance of the correlation function; a covariance correlation function based color cast detecting algorithm is designed corresponded to stable pixel correlation in casted video.(3)the system is implemented on the VS2012 platform with C++ language and Opencv image processing method. The whole framework is built with MFC and tested with various videos and pictures. The results indicates that the proposed system is applicable for real-time detection with high precision and stable performance.Finally, this paper made a summary of wok done, and a prediction of the application of real-time monitoring system. monitoring real-time detection system were discussed.