Study On Digital Image Inpainting

Digital image inpainting is a technology to deal with image, photo or film, and let their injured region renewed, removed or replaced the objects chosen in the picture, which is an important branch of digital image restoring technology. The work principle of it is to use the around information of inpainting region or the former and last frame information of the film fill in the region. It mainly been used in the image transferring, image compressing, or inpainting injured images, hiding aims and so on. With the development of computer, how to do image inpainting automate or half-automate on the base of digital image becomes an important task in the field of image inpainting research.The paper firstly introduces two image-inpainting methods based on total variation and on texture synthesis, analyzes their advantages and defects, uses vector quantization etc. optimize the method on texture synthesis. To make the best of the both methods' advantages and escape the defects, an image decomposition model is introduced. Firstly, images aredecomposed the texture part and the structure part, which keeps edges and smooth, then the image-inpainting methods based on total variation and on texture synthesis are used in the corresponding parts, finally the last result is composed of the two finished parts. From the experiments analysis, the image-inpainting method based on the image decomposition achieves good results, especially in the region that include both texture and structure. The study does a favorable basis on the next research work.