Digital Video Image Processing Based On The Concept Of Adaptive Neighborhood Statistics

Recently, with the rapid growth and development of video applications, video services and consumer electronics, video image processing comes to be a new hot topic in multi-media field.Traditional image processing has been comparatively mature, but big difference exists between the video image processing and the traditional image processing in many aspects mainly including the issues need to be resolved, the existing difficulty need to be face with, and the background and environments need to be applied to. So fast, highly effective, and reliable algorithms special for video image processing has been a research topic with great significance and value in theory and practical application.Proposal of the concept of adaptive neighborhood provides a simple but effective method for image processing based on object division. Because of its significant advantage in different application background it has got widely research and application. However many problems still exist and need to be further researched and resolved when using it in video image processing.This dissertation focuses on the key issues about how to introduce the concept of adaptive neighborhood to video image processing, make use of its advantages, conquer its disadvantages, and finally use it to resolve the problems in area of video image processing. Main contents and contributions of this thesis are listed here:1. Briefly reviews the history and development of digital video image processing and summarizes the main topics of it. Emphasizes on the introduction of the generation and development of adaptive neighborhood and the start-of-art of it. Based on the detailed analysis on the existing results in this field indicates the advantages and disadvantages of adaptive neighborhood when using it in image and video image processing.2. Proposes a differential video noise estimation that uses spatial and temporal relativity of video signal. Adequately usage of temporal relativity special for video sequence results in the improvement of reliability. The results of simulation and application in practical system show that the proposed method performs well in both highly noisy and good-quality videos.3. Researches on an effective video image pre-processing algorithm. Based on the analysis of true situation defines the design objectives and requirements for a successful video image pre-processing. Then proposes a video image pre-processing method based on adaptive neighborhood statistics. The proposed pre-processing conquers the low efficiency of classical adaptive neighborhood filter and the sensitivity to threshold in other methods. It can restrain the spatial and temporal noise without blurring the edge and detail information in videos which has got proved in simulation and practical application.4. Researches on a fast and reliable algorithm to restrain salt-and-pepper noise with high density. Proposes the concept of Stepwise strategy and neighborhood priority in the repair processing. Because the proposed method integrates the advantages of the median-filter and PDE methods, it can restrain the salt-and-pepper noise for the noise density range from 10% to 95%, and at the same time effectively repair the edge and detail information in videos. Moreover proposes some speed-up design and optimization strategy which promotes the efficiency significantly.5. Proposes an intelligent exposure correction solution based on two scales video images decomposition using adaptive neighborhood filter. It overcomes the shortcoming of existing solutions, and proposes giving different exposure correction to the two scales video images decomposed by adaptive neighborhood filter which can avoid non-linearly magnifying the noise and make it can be applied in the scene of low lighting, high lighting, and the more troublesome situation when both of them exist synchronously. Also it satisfies the requirements of real-time application and has been used in practical digital entertainment products.6. Researches on the effective method to estimate and restrain the blocking noise in video coding system. Proposes a new strategy that immediately estimates error image in encoder end and uses the result of it to help de-blocking filter in decoder end to restrain the blocking effect. The new architecture improves the veracity of blocking noise estimation and the performance of de-noising. The results of simulation show that the proposed algorithm can not only give reliable estimation for error noise of video coding in encoder end, but also restrain the blocking artifacts in the decoder end without blurring the edge and detail information.