Study Of Variational Models And Algorithms For Retinex

In this paper,we propose two variational models for Retinex-based image enhancement.In the Retinex-theory-based variational models for image enhancement,logarithmic transformation is usually used as preprocessing.The advantage of this operation is that the multiplicative model becomes an additive one,which makes it easier to model and design the algorithm.However,this operation will result in the loss of details in reflectance.To overcome this shortcoming,in Chapter 2,we propose a detail preserving variational model for Retinex to simultaneously estimate the illumination and the reflectance from an observed image.Different from the logarithmic-transformation-based models,the proposed model performs the decomposition directly in the image domain.Theoretically,we prove the existence of a solution for the proposed model.Numerically,we derive an efficient iterative algorithm by utilizing the alternating direction method of multipliers(ADMM)method for it.Experimental results demonstrate that the proposed model can overcome the loss of fine details efficiently to gain enhanced results of high quality.In Chapter 3,based on the analysis of previous methods of image enhancement,we propose a Retinex-based fractional-order variational model to better decompose an observed image for night image enhancement.The proposed model can get details apart from dark backgrounds to get better results.Similarly,we utilize the alternating direction method of multipliers(ADMM)method to solve the model.Compared with other closely related Retinex methods,the proposed method achieves competitive results on both subjective and objective assessments.Besides,we also test the model on the other two image datasets which shows that the proposed model has a good generalization.