Research On Theory And Key Technologies Of Color Image Enhancement Based On Human Visual Perception

As imaging technology develops,image enhancement is an ideal processing method in many areas,such as scientific research,military application,forest disaster,agricultural yield estimation,tracking and petroleum exploration,laiding the foundation for high-quality images for subsequent processing.However,the limitations of the camera own dynamic range,severe weather conditions,and non-uniform light during the imaging process will cause image blurring and visual effect reduction.The thesis focuses on the theory and applied research of color image enhancement for human visual perception,which has important theoretical significance and application value.Based on the review of the research status at home and abroad,the main work carried out and the innovative results achieved are as follows:1.After analyzing the cause of detail blurring and halo noise,a novel hybrid tone mapping method which combines global and local tone mapping is proposed.First,the captured image is processed by global tone mapping which preserves the global lightness and contrast;Second,local contrast and details are captured by Sigmoid based local tone mapping method.Finally,the bilateral filtering is brought into processed reslut to suppress halo noise and preserve edge details.Experimental results demonstrate that the proposed algorithm not only enhances global lightness and contrast,supreess dynamic range,but also prevent halo artifacts.2.Surround suppression based variantional Retinex is proposed.Instead of a gradient module,a surround suppression mechanism is used to provide spatial information in order to adaptively distinguish contour boundary and texture edge information.The proposed strategy preserves the boundary areas in the illumination so that halo artifacts are prevented.In addition,in the smooth area,stronger regularization constrain is used to eliminate nonuniform illumination effects.The split Bregman optimization algorithm is employed to solve the proposed model.Finally,after decomposition,a Laplacian-based gamma correction is added to illumination for prevent color cast.The recombination of the modified reflectance and illumination become the final result.Experimental results demonstrate that the proposed algorithm performs better than other methods.3.A new joint regularization method considering image degradation and naturalness preservation is proposed.First,an anti-degraded model named dark channel prior(DCP)is introduced into the captured image to get haze-free image.Second,the haze-free image will be decomposed into detail and naturalness components by light filtering.Then,a logarithmic Laplacian-based gamma correction(LLGC)is added to naturalness component for preventing color cast and uneven lighting.In addition,the associated reflectance component will be obtained by using proposed variational framework according to the hypothetical regularization priors.Finally,the recombination of associated reflectance and naturalness component become the final result.Experiments show that the proposed joint regularized variational Retinex algorithm achieves the purpose of removing environmental degradation and enhancing the image,and suppresses color cast and halo artifacts.