Application And Research Of Image Sharpness Detection Based On Region Edge Model

With the development of society,video surveillance system has been used in people's production and life widely and deeply,especially in government,education,transportation,finance and other important areas and places.It will be very difficult to make sure that the video is clear and normal faced so many and widely used cameras.Therefore,we have studied and analyzed the videos with abnormal sharpness,so that the system can alarm it automatically before the accident occurred and reduce the losses.To test the problem of quality degradation of security monitoring video caused by improper camera parameter setting,a video sharpness testing method based on the edge direction of the optimal straight-line segment is proposed.In view of the characteristics of overall video quality degradation caused by the blurring of the surveillance video image and the totally different performances in local areas,an accurate video image sharpness metrics can be obtained through the extraction of the optimal straight-line segment and the optimal edge direction,combining with the strong edges of the image.The algorithm is tested in the LIVE database and security monitoring video database,and the correlation analysis is also conducted.The experimental results show that the video evaluation data obtained through this method,has good consistency with the subjective assessment,meeting the actual needs of the security monitoring.The edge-based non-reference image sharpness evaluation method has a good consistency,but later experiments,found that for serious abnormal sharpness video,the algorithm is not very good to adapt to.Therefore,aiming at this problem,combining the spatial domain and the frequency domain,an evaluation method based on edge and energy is proposed.This method introduces multi resolution decomposition on the basis of the edge in the spatial domain that can improve the sharpness of the image in the serious images.In the frequency domain,the image is decomposed by wavelet,and then all the high frequency parts of each layer are merged into a detail map,and the energy of each layer is calculated by using its gradient,then the energy of all layers is combined to obtain the high frequency energy of the whole image.Finally,the sharpness of the image is evaluated by using the edge width in spatial domain and energy of the high frequency in frequency domain.Experiments show that the method has a good consistency and the performance is stable.Finally,the related work of this paper is summarized and the future research directions are pointed out.