Image Enhancement Based On Partial Differential Equations

Reality, image was polluted in the process of transmission and forming by digital imaging devices and the external environment.The process of reducing the noise in the digital is called image denoising. Image denoising based on PDE is a good image denoising technique, there are two big mainstream model, process oriented, such as PM, FAB and TKY model; Another kind is object oriented, such as variational denoising model. But the use of a single one defect of PDE denoising method are:1) the first kind of method is often not very good in smoothly or sharpening regional noise information and image information, make some image information as the denoising process to eliminate; 2) the second method is calculated on a flat area of the false edge have a staircase effect.Aiming at this problem, this paper presents a combination of two types of PDE model denoising method. On the basis of the variational model combining, PM, FAB method can not only keep the important image information, also avoid the ladder after denoising effect. The algorithm is based on noise parameters determine the fidelity term of variational denoising model, has reached to protect the image information, and for the model regularization is combined with PM or FAB diffusion parameters to remove the noise, smooth image. In this paper, the whole process of concrete steps are:1) do to contaminated image based on the histogram of the noise type identification and parameter estimation; 2) according to the different frequency histogram curves to determine noise type, similar to salt and pepper noise random noise, the variational denoising can not only remove very well. If judgment is the gaussian noise and speckle noise, so you also need to estimate the noise variance, according to the variance to determine the variational model constraint conditions, so as to remove noise.In this paper the method based on the histogram analysis has reached the estimate of the noise type and parameters, and according to estimate the parameters using the corresponding method to filter noise. Through comparing with the traditional strengthening methods, which was not only verified for gaussian noise that the method based on PDE was better, but also verified for speckle noise, the method of combination enhancement effect was superior to combined with the former method. Finally, summarized the partial differential method needs to be improved, and the emphasis and direction for the future research prospects.