Research On Image Definition Evaluation Algorithm For Video Surveillance

Along with the monitoring and control system for all walks of life,to popularize and apply in the system to accommodate the number of cameras in gradually increasing,which increases the difficulty of system maintenance,some potential problems will be revealed,to the quality of video detection is one of them.In video monitoring system,a single camera transmission picture sometimes abnormal,for example,camera video signal,snowflake point in partial color or transmit images show that the more serious effect on the quality of video image,these conditions can affect the monitoring effect of video monitoring system,if not timely find and solve these problems,may cause certain economic loss.However,only by artificial detection methods to detect the camera whether the video image transmission by the abnormal is a very troublesome thing,workload is big,complex,both time-consuming and laborious,is not conducive to reduce the cost.Therefore,let the system real time automatic detection video image quality of camera is necessary.In a wide range of video quality evaluation index,the video resolution is an important indicator of video quality.Apply video clarity evaluation to monitor system,which helps the system maintenance personnel timely understanding of the operation situation of the camera,such as whether to correct,block,or heavy focus,can effectively improve the efficiency of equipment maintenance.In recent years,the researchers according to the edge of the point spread function(PSF)and image to estimate the sharpness of the image,based on the analysis of the human visual system(HVS)characteristics of contrast sensitivity,using multilevel wavelet transform to extract the key factors affecting the quality of image resolution,through the study of the HVS characteristics of the key high frequency energy weighted,the construction of a single frame image clarity evaluation index.On this basis,the similarity is proposed based on frequency domain of video surveillance sequences,key frame extraction algorithm through weighted keyframes articulation index,the definition of the video sequence evaluation algorithm was established.And to simulate the algorithm and the subjective evaluation of comparison,and analyzes the performance.In this paper,the main research work is as follows:(1)in video surveillance images,and put forward the key frame extraction algorithm based on frequency domain similarity.The single frame image low-pass filter in the first place,remove the noise.Because the video screen there is strong correlation between adjacent frames,in the same scenario,the numerical changes between adjacent frames is lesser,a layer after wavelet decomposition,to use the current frame after the wavelet decomposition of low frequency image relative to the previous frame wavelet decomposition have low frequency images without change,change size to distinguish between the key frames and scenes.(2)in view of the key frames,is proposed based on HVS characteristics and the wavelet decomposition of the single frame image clarity evaluation algorithm.First for 2 d multi-level wavelet decomposition,according to each of the decomposed wavelet component,extraction of the corresponding energy spectrum,the weighted HVS characteristics contrast sensitive characteristics,and then to calculate the energy of the high frequency component of the image after the weighted ratio,as a key frame image sharpness evaluation index.And the single frame image was unimodal,noise and human eye related to performance evaluation.(3)of the above algorithm simulation verification.For key frame extraction from the fidelity and compression ratio for performance analysis.For image and video resolution from the unimodal,noise and visual correlation for performance analysis.