Font Size: a A A

Video Techniques Based On Computer Graphics Algorithm

Posted on:2008-06-09Degree:MasterType:Thesis
Country:ChinaCandidate:L YangFull Text:PDF
GTID:2178360242476852Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
Video compression in source coding has marched forward a long step with H.264/AVC standard, in which a variety of advanced technical tools have been presented to reduce spatio-temporal redundancy. However, the highly compressed bitstreams would be induced greater sensibility to the channel disturbance. Consequently, error recovery combining either active error resilience or passive error concealment technique, dedicated to minimizing video quality deterioration by inherent texture spatio-temporal correlation, is especially demanded. On the other hand, great advancements have happened in computer graphics and vision domain. Motivated by the plausible substitute for the unknown regions in image inpainting, it has been applied to lossy image transmission. Thus, it is a very hot topic to apply computer graphics and vision techniques to generate a plausible substitute for the unknown regions with complex structure while errors occur in error-prone video transmission.As one of the most important features of structure infromation, curvature-based structure representation of image could be considered in a multiscale sense. Thus, an adaptive error resilience algorithm using redundant shape coding with multiscale B-spline based feature localization is proposed in this paper. In the proposed scheme, the curvature-based curve representation information extracted from original video pictures is coded and encapsulated into redundant slice, a new error resilience tool adopted into the H.264/AVC standard. If the primary slice is not the redundant slice would be used to provide a low level of reconstructed quality with high-level feature based video completion algorithms.Furthermore, for video sequences suffering from data loss during transmission, we aim to complete the missing parts in such a way that those target texture could not only preserve the complex and semantic structure, but also ensure both maximum tempo-spatial consistency and less error propagation. In this paper, we propose a video completion scheme embedded in a universal video decoder, where the global optimization problem is formulated into a local optimization process of available variables as confidence propagation. In consequence, the optimal solution is attained by each frame to minimize global visual artifacts. The proposed local optimization with confidence propagation would be decomposed into: the objective temporal patch matching, and the inferred spatial patch orientation, followed by spatial completion algorithm. Instead of traditional spatial completion algorithm in error recovery mechanisms, our paper proposed a spatial completion algorithm aiming to reconstruct the unknown regions in lost packets utilizing inherent correlation between the lost blocks and its neighbors. For regions with complex geometric property, a structure-driven directional interpolation is proposed to restore information in the unknown region near structure. Thereafter, a directional patch placement for Graph Cuts is dedicated to achieving uniform texture within unknown regions. To offer a best-first filling between structure continuity and texture confidence, a filling priority compensation is devoted to further enhancing restoration results.
Keywords/Search Tags:H.264/AVC, video completion, structure-based, B-spline, inpainting
PDF Full Text Request
Related items