Research On Algorithms For Fractal Image Coding Technology

With the progress and development of the digital technology and network, the digital image technology has been widely used in computer network, digital cameras, digital cameras, digital televisions and mobile phone. Therefore, images in digital form for storage, transmission and sharing are the inevitable achievements of technology development. The image can be filtering, transform and synthesis with other images when stored them in digital form, so, sometimes there will be a lot of factors which will defect the information of the image for some purpose, when the image is transform in digital form, such as remove specific objects of the image, modify the image contents, add some extra information to original image and so on. To deal with the above problems, the digital image restoration techniques emerged. Digital image inpainting technology refers to use the known information of the image and a specific method to repair the damaged image region, such that the restored image is as close as the original image. The main content of this thesis is the use of fractal coding method to digital image restoration.This thesis introduces two kinds of digital image restoration algorithm by fractal coding. The first restoration algorithm is based on the iterative fractal decoding and edge extension, the second restoration algorithm is based on the double fractal code decoding and adaptive TV algorithm. The principle of Fractal Decoding is using a initial image and the fractal information of original image to approximate the original image. In fractal decoding, we use an initial image and the iterative function to generate a decoded image to approximation the original image. In the image restoration process, we hope that the initial is very similar to the original image. But the initial image fractal decoding is selected randomly, so the similarity between the initial and the original image is often very small. The method based on the iterative fractal decoding and edge extended made an improvement to the usually digital image restoration algorithm base on fractal code, it uses the restored image as the initial image for the next iterative fractal decoding and image restoration. In order to ensure the image block and the fractal storage block not be tampered together, the algorithm use two push operations and an exchange operation to obtain the index table which can dispersed storage location as soon as possible. At the same time, this algorithm only needs the last two significant bits to store information, which can improve the signal to noise ratio of the embedded image.Another image restoration algorithm is based on double fractal code decoding and adaptive TV algorithm. In fractal decoding, first of all, we use the received fractal code about the contour information and adaptive TV algorithm to restore the main structure of the original image, followed by the use of fractal code decoding to extract detailed information secondary recovery steps to repair the image. Then, we use another fractal code which contains the detail about the original image to repair the tampered image again. The process between the first fractal decoding repair and the second process is basically the same, the difference is that, in the first iterative decoding, the initial image for fractal decoding is selected randomly; however, the initial image in the second part of the decoding is generated by the first part of image repair. During the next iteration, the initial image of the first part of the fractal decoding is the image generated by the last iteration of repair. Firstly, the original image is sampled to generate a sample image, and then extract the fractal code of the original image and the sampling image. We first obtain the fractal code of the middle position of the sampling image use the loop form from inside to outside; because we default think that the most important content of the image is in the middle, so as the fractal code of the original image. Combine these two sub-codes; we can get a fractal code which both contain the contour and details information of the original image. And then according to the generated index table, we embed the fractal code into the original image. Finally, embedding the Wong watermark into the LSB1bits of the image to obtain the watermark image. In image recovery, algorithm first use the Wong watermark to locate the tamped region, and then iterative use extracted fractal code of the sampling image and the adaptive TV algorithm to restore the tampered image to generate a primary repair image which contain the contour information of the original image, finally, we iterative use fractal information which contains the details of the original image for the last repair. This thesis proposed two kinds of image restoration algorithm through a lot of experiments, and the experimental results show that the two image restoration algorithms are validate.