Based On Video Traffic Detection System Research Of Image Enhancement

In the intelligent traffic system, such as accident management system, road toll system, parking management systems, global positioning system, first of all need to obtain the sections or bayonet HD camera, then a full or partial information identification, monitoring and positioning. But due to the equipment and weather constraints, it is necessary to enhance the image, the image identification to ensure correctness and finally the effectiveness of regulation.Based on the background of engineering application, this paper focuses on the intelligent transportation system in the original image is a series of enhanced processing, and then the image quality evaluation, so the original image after the algorithm enhanced, has very high recognition, in the follow-up image processing, has a higher recognition.Firstly, based on the original image histogram equalization, judge the original image for low contrast image, for the low contrast image, into the algorithm was enhanced; if not, it does not require the original image enhancement processing.Secondly, the improved CLAHE algorithm to the image processing, obtains the enhanced image. Improved CLAHE algorithm for image enhancement, the original image space transformation from RGB to HIS space, the use of HIS color model, color quantization is constructed by using the HIS space code templates, simple calculation of I space images, contrast enhancement, reached the final of the purpose of enhancing image.Once again, to enhance the image wavelet denoising, further removing image noise. Wavelet denoising is because the wavelet has the characteristic of multi-resolution analysis, the image signal is decomposed in different scales, and also can be mixed signal is decomposed into different frequency bands of different seed signal, the image signal frequency division processing function. This processed image can be more effective in removing noise.Finally, through the wavelet denoising, image information will become more smooth, edge information is particularly obvious, is not conducive to the recognition information details. In this paper, using cellular neural networks for image edge sharpening. The experimental results show that, using CNN to extract the edge directions, and in the other the classical edge detection operator fails, CNN can still be successful.