| In foggy days,the images which the image acquisition system get have the problem of blurring,color distortion and contrast reduction because of the scattering effect of the tiny particles in the atmosphere.This not only influences the visual effect of the image,but also interferences the image feature extraction,which makes automobile visual system running severely reduced and pose a security risk.Therefore,improve the effect of smog day image haze-remove algorithm quickly and effectively has a practical and theory of urgent needs.Main for the fog day degradation problem of the automobile visual system collecting images,this paper introduced the key technology of automobile visual system image processing method and image degradation reasons and fuzzy mechanism in fog days,analyzed the mechanism of haze-remove algorithm based on dark channel prior principle in detail,and made a deeply research on the advantages and disadvantages of guided filter and bilateral filter improved dark channel prior algorithm.The paper offered a new method based on the application needs of mobile visual system,which further improved and perfectly presented the above two improved black channel prior image haze-remove algorithms,specific contents are as follows:1.In order to solve low contrast ratio and color distortion problems,the method of equally spaced sampling and brightness mutation removal had been used to estimate the atmospheric intensity value of the image and sub-region method had been used to optimize calculation of transmission rate;2.Aimed at the real-time requests of the mobile visual system,bilateral filtering method had been used to instead "soft matting" for feathering transmission rate to improve the processing speed;3.According to different fog densities have different effects on images,this paper use three different algorithms of newly proposed improving method of this paper,bilateral filtering and global filtering based dark channel prior image haze-remove to separately process dense fog,moderate fog and shallow fog in order to get a clear image.Experiments show that this method has solved the real-time and the color distortion problems of the automobile visual image in foggy days and increased the operating results. |
