Video Communications, Image Restoration Model And Algorithm Research

As human vision plays an important role in perceiving the outer world, the needs of visual information are constantly increasing in recent two decades, which greatly impels the vision related science and technology advance. Any applications concerning human communication can be improved through the introduction of vision, thus technology on image or video communication is developing fast.Many image coding algorithms have been developed to reduce the bit rate for digital image and video representation or transmission. Among them, block-based techniques have been proved to be the most efficient one and thereby are used in most existing image and video compression standards such as JPEG, H.26x and MPEG-x. Since real-world communication channels are commonly not error free, corrupt of data transmitted through such channels is generally unavoidable. Moreover, block-based image coding systems are vulnerable to transmission impairment. Loss of a single bit often results in the loss of a whole block and even may cause consecutive block losses consequently. Therefore, proper methods for error resilience must be applied in order to obtain good restoration quality of images.In this dissertation, some issues on the techniques of error resilience and image inpainting are discussed, which are detailed as follows:(1) Both domestic and overseas research status in quo and typical algorithms of error resilience in video communication are investigated, whose performance are analyzed, too. Among them, Best Neighborhood Matching (BNM), an error concealment algorithm with good performance of image restoration, shows pretios. However, BNM finds matching blocks in way of full search strategy that needs intensive computation, which restricts its practical usage, especially in the scenario of real-time image processing. Fast block-matching algorithms are therefore emerged and according to search strategy, three novel fast image restoration algorithms are proposed, namely Two-Step BNM, Diamond Search BNM, and Square-Diamond Search BNM. The significant performance improvement is proved by simulation and experiments.(2)Image inpainting algorithms are studied hereof. A new error concealment method in video communication is developed, which employs ideas of partial differential equation. Being viewed as an optimization problem in terms of discrete energy function, the proposed method herein is derived from an image inpainting model proposed by Tony Chen. Using the neighborhood system of Markov field, we have presented a digital total variation model forerror block concealment and induced the corresponding numerical iteration scheme. Incorporating with spider network scan technique, a simple and fast algorithm is designed and implemented. Our algorithm can ensure the continuity of the edges and has been applied with good results in the transmission of video data over non-reliable packet networks.