Research And Application Of Patch - Based PDE Image Repair Model

Image inpainting is an important research branch in image processing. It is aimed to fill in the information for the damaged regions and to restore the integrity of the image by using the information around. There are many applications in protection of cultural relics, restoring aged or damaged photographs and films, text removal, and so on. In this paper, we focus on the variational PDE models for non-texture image inpainting. Moreover, some theoretical analysis about them and the corresponding numerical algorithm to realize them are given.In this paper, we have a comprehensive analysis and introduction for BSCB inpainting model, TV inpainting model, CDD inpainting model and Euler’s elastic inpainting model in basic inpainting principle, model construction, and the advantages and disadvantages of each model. In the light of some problems existed in these models, more improvements are made and two new inpainting models are proposed as follows:1.We propose a novel CDD inpainting model by introducing a parameter to the traditional CDD inpainting model, then its simulation implementation are realized by using the MATLAB software. As can be seen from the experimental results, the inpainting effect of the novel model is improved to some extent compared with the original model. In order to illustrate the feasibility and practicability of the proposed model, we also effectively apply it to the field of restoring aged photographs, protection of cultural relics and special effects. What’s more, the numerical simulation shows the superiority of the novel model.2. We introduce the term｜▽u｜p(0<p<1) instead of the total variation in the energy functional and obtain an improved model for Euler’s elastic inpainting model. Furthermore, it is analyzed and proved in theory. Accordingly, the staircase effect due to the total variation is reduced effectively. In addition, peak signal to noise ratio (PSNR) and structural similarity index (SSIM) are used to evaluate the image inpainting quantitatively. Experiments and comparisons show that values of PSNR and SSIM acquired by the proposed model are much higher than the other models’; The inpainting effect is improved obviously. Compared with the traditional Euler’s elastic model, the effect of the new model is much better, which is consistent with the needs of human visual psychology.