Research On De-noising Method Of Vehicle Detection

With the development of science technique that makes great changes of information society, the multi-media technique becomes an important part of computer development in 21st centuries. People through various medium such as language, music, writing, graphics, image, television to carry on the processing of information, deliver and save. But usually the noise and image keep both, the existence of the noise will make big influence to the processing of information, deliver and save. Therefore, how better reduce the noise, is a classic problem of the image processing fields.Noise will inhibit people's comprehension to an image. The aim of de-noising is removing noise in the image, improving the degree of people's understanding to the image, and processing the image conveniently. The image de-noising method basically may divide into two kinds: transform domain de-noising and spatial domain de-noising. The former reduces the noise through three steps: first, transforms the image from the spatial domain to the transform domain by some kind of transformation. Second, deals with the transformation radio in the transform domain. Last, transforms the image form the transform domain to the spatial domain by inverse transformation. The latter carries on the processing of the gray value of each pixel directly at the two-dimensional space to which the image belongs.The application of de-noising is extremely widespread, appears nearly in all image processing domains. For example: in the vehicle detection system, we deal with the fixed-camera video image, extract the moving object and get the vehicle and the background division. But, the image which we get now contains much noise in either background or vehicle. It needs de-noising to this image to gain a more quality image. The main task of this paper is to propose a new binary image de-noising method aims at the existing noise question after the vehicle detection. This method can detect and reduce noise in image, get moving object division effectively and integrally. Experiment shows that this method can get better performance, makes preparations for the following shadow detection, vehicle recognition, vehicle tracing, and has good robustness and timeliness.