Research On Enhancing Technology Of Low-quality Highway Traffic Images

In recent years,the impact of bad weather such as haze,rain and dust storms has led to a serious decline in the quality of road traffic pictures taken by highway traffic monitoring and automotive electronic driving recorders.This paper focuses on low-quality road traffic images,mainly including road traffic images under fog,rain,and dust storms,to enhance algorithm research,select prominent and typical low-quality images as research objects,for image blurring,low contrast and color Saturation is not high,the local information is not obvious and other issues are compared.The main research contents of this article are as follows:(1)For the analogy research based on low-quality image enhancement technology,this paper studies an improved histogram enhancement algorithm,which divides the maximum entropy value in the component information,achieves the improvement of image spatial enhancement and improves the contrast of the image.Secondly,the results of the traditional dark channel image enhancement methods are dimmed and prone to slight halos.This paper studies an improved dark channel enhancement algorithm.This method captures the parameters such as the atmospheric light value and the transmittance of the image,and uses the spatial component information and the boundary component information of the image for optimal processing.(2)Concerning the traditional contour wave Contourlet algorithm to produce over noise and deal with the deficiencies in details.This paper proposes a Contourlet algorithm based on "Post-wavelet" and a multi-scale enhancement algorithm in Retinex algorithm.(Multi Scale Retinex.MSR)combined image enhancement algorithm.After the initial decomposition of the Contourlet,the algorithm combines the MSR algorithm for the low-frequency part containing the main information of the picture,and then reconstructs the low-frequency information and high-frequency information to finally complete the enhancement of low-quality images.(3)On this basis,the Gaussian function in the low-frequency information processing MSR algorithm is improved to be a guided filter function,and the local mean value and variance of each pixel in the neighborhood are used for local estimation processing to enhance the detail processing of the image.Using the basis function of the multidimensional signal,the resolution of the image is continuously approached with the scale transformation.The enhancement result has more prominent edge details,and the sky area recovers brighter,which can display the detailed information of the image and the overall contour.Finally,by analyzing the vertical and horizontal comparisons between the improved algorithm and the low-quality image enhancement analogy method,we can see that the improved algorithm can achieve good image edge enhancement,color saturation,close-range information recovery and other three aspects effect.The results of the algorithm in the haze world of low-quality road traffic images are relatively good.The enhancement results have more prominent edge details.As a result,the information entropy of the images is relatively high,and the main information of the images can be better recovered.The enhancement effect is clear,and the detail texture processing is also superior to the improved dark channel algorithm.Compared with the histogram improvement method,the noise of the image is also well suppressed.