Research Of The Image Compression Based On Partial Differential Equation Inpainting

Image inpainting, the technique which is used to restore the damaged calligraphy, paintings and old photographs, or remove and replace the selected objects, is as ancient as art itself. In 2000, the inpainting model based a three order partial differential equation (PDE) was first introduced into the image processing in the paper of"image inpainting"by Marcelo Bertalmio, etc. This method is automatically used to those damaged regions by computer, and marks the normal presentation of the digital image inpainting.The combined application of image inpainting based on PDE and image compression is mainly researched in this thesis. Firstly it presents the advantages of PDE in image processing as an analysis tool, and its special numerical solution. Then the detailed techniques and simulation experiments are introduced, as well as its advantages and disadvantages. After the study of PDE and the analysis of PDE inpainting, we bring a new quick image preprocess, and achieve a new image compression with low data and complication under the corresponding encode and decode method. By comparing with normal compression without PDE inpainting, it proves that the PSNR is improved, and the block effect at a low code rate is reduced efficiently in our method. That means new method is feasible and superior.