Total Variation Based Image Quantization Noise Removal

With the development of Internet technology, people have a growing demand for the information, and the information is mainly viewed as image. In order to decrease the storage space of the compressed image and save the transmission time, we have to compress the images. Digital image compression is an important aspect of the digital image processing.JPEG has become an important standard for image compression. It can select the image compression ratio freely. Since we round the data in the quantization process, it results in losing some image data and thus has formed the quantization noise, which leads to the well-known JPEG blocking artifacts. The larger the compression ratio is, the poorer the results of decompression images are. Then we can see the blocking artifacts obviously. On the contrary, when the compression ratio is small, we get a better decompression image. During the process of information transmission, for saving the processing time, there has been one way to control the storage space of images. When the storage space has satisfied a certain standard, we will get a poor image that has low quality. In this paper, we discuss how to eliminate quantization noise and reduce blocking artifacts, which can improve the quality of decompression image. We use the total variation method to discuss the further improvement in the gray scale and color images compression applications. The main work in this paper is as follows:(1) We review the researches in recent years about how to eliminate the image quantization noise in image compression. In this paper, we introduce the standard of JPEG image compression and the causes of blocking artifacts.(2) We introduce the basic theory of total variation method and apply the TV model to remove the quantization noise. We rewrite the minimization problem to a saddle-point problem. We use the primal-dual algorithm to solve the optimal problem. Last, we simulate the results of experiment.(3) Because the problem that removing pieces of image information in the quantization process of the TV model, such as removing details and texture, we propose a new model based on non-local total variation. We discuss the basic theory of the new model and apply this model to the gray scale images. The experimental results show the superiority of this model.(4) We set up a color image compression model and apply the total variation method to the color images. We review the basic theory of color TV model. The experiment shows the superiority of this model.