The Blurred Image Restoration In Intelligent Transportation System

In recent years,many countries are facing with various traffic problems,such as the serious traffic jam,frequent traffic accidents and so on.In order to solve these problems,advanced science and technologies(such as communication technology and network technology)are effectively applied to the entire transportation management system,namely intelligent transportation system(ITS).Image processing technology in the field of ITS application is very widely,and traffic monitoring system based on video or image is one of the main way to obtain the traffic information.In the process of image information acquisition,due to the relative movement of the camera and the object or the optical system lens out of focus and so on,blurred phenomenons appear in the image.Blurred image restoration is an important research field of digital image processing,and motion blur and defocus blur are common factors which lead to blurred images.Point spread function of the parameters estimation is the premise and key in motion blur and defocus blurred image restoration,which is the focus in this paper.For estimation of parameters in motion blur image restoration,its spectrum and cepstrum features are analyzed,and automatic detection based on frequency and cepstrum algorithm is proposed.For estimation of blurred angle,the Radon transform algorithm based on frequency of the blurred image is used,which can accurately detect the blurred angle.For estimation of blurred length,the differential curve of frequency projection algorithm to get minimum value is used in the paper,which can accurately detect blurred length.When the point spread function during motion degradation is estimated,motion blurred image restoration can be realized.Experimental results show that,without the noise,in the 15-75 pixel,error less than 0.2 pixel.For estimation of parameters in defocus blurred image restoration,Autocorrelation function of Laplace is used to identify defocus radius in the paper,Firstly,differential autocorrelation function of the blurred image is calculated by the Laplace operator.Accordiry to the ring grave from projection of differential autocorrelation function,the bottom of thegroove become into the differential circle,whose center is zero frequency correlation peak,and the required radius is the double of the defocus radius.Experiments show that the algorithm noise resistance is strong,and the radius error within 0.5 pixel.For both degraded image restoration,inverse filter,wiener filtering,least square and Lucy-Richardson algorithm and improved Wiener filter is used to achieve image restoration in the paper,and the improved wiener filter and Lucy-Richardson algorithm work better.