Research On Low-Light Image Enhancement Algorithm Based On Illumination Estimation And Camera Response Model

Due to insufficient light at night and imperfect acquisition equipment,the obtained images generally have poor visual effects,low brightness,serious color distortion,large noise and loss of details.So it's bad for human observation and computer vision algorithms.To solve this problem,many image enhancement techniques have been proposed so far.However,existing technologies inevitably cause color and brightness distortion when increasing visibility.In order to reduce the image brightness and color distortion,the existing low illumination enhancement methods are studied in this paper.A novel low illumination enhancement using light estimation and camera response model is improved.The main research contents of this paper are as follows:In this paper,based on dark channel prior theory,the pixel value of low illumination image is inverted to obtain the defogging image.The transmittance and atmospheric light can be obtained by using dark channel prior.Then the dark primary color is obtained adaptively and the guided filter is applied to the transmittance to get a more real transmittance.Finally,the transmittance is improved to obtain the image with enhanced low illumination.A low illumination enhancement algorithm based on light estimation and camera response model is proposed for images with extremely low brightness.The processing inside the camera is taken into account when designing the low illumination enhancement algorithm.For most digital cameras,the pixel value is not directly proportional to the amount of illumination falling on the camera.The nonlinear function related to camera sensor illumination and image pixel value is called camera response function(Camera Reponse Function,CRF).The mapping function between two images with different exposures taken in the same scene is called the luminance conversion function(Brightness Conversion Function,BCF).In this paper,a new enhancement algorithm is presented,which takes CRF and BCF into account.An accurate camera response model is obtained by combining CRF and BCF.There are two key issues involved: one is to find the appropriate camera response model,and the other is to determine the light map and get the exposure rate map.Finally,the illumination diagram is used to estimate the exposure rate diagram.The illumination map is obtained by weight estimation and comparison of several methods.Gamma correction is performed after the illumination map is obtained.In this case,the exposure rate graph is the reciprocal of the illumination graph.In order to avoid the illumination graph being zero,parameters are introduced.The enhanced image is then obtained by using the camera response model to adjust each pixel to the desired exposure based on the estimated exposure rate diagram.In this paper,a number of experiments are carried out to verify that,compared with several commonly used enhancement methods,the enhancement method based on light estimation and camera response model can reduce image color and brightness distortion,and the time and space complexity of the algorithm is low.Therefore,it can be applied to practical work and has good practicability.