High Efficiency Video coding (HEVC), not only provides a much better coding efficiency than previous video coding standards, but also shows significantly superior performance than other image coding schemes when appli...
详细信息
ISBN:
(纸本)9781479902880
High Efficiency Video coding (HEVC), not only provides a much better coding efficiency than previous video coding standards, but also shows significantly superior performance than other image coding schemes when applied to image coding. However, the improvement is at the cost of significant increase of encoding complexity. In this paper, we focus on retaining the high coding efficiency provided by HEVC while largely reducing its encoding complexity for image coding. By applying various techniques including optimized coding structure parameters, coding unit early termination, fast intra prediction and transform skip mode decision, we significantly reduce the complexity of HEVC intracoding while keeping most of its coding efficiency. Experimental results show that our light-weight HEVC encoder can save about 82% coding time compared with original HEVC encoder. With a slight loss to the HEVC reference software, the proposed scheme still gains about 19% in BD-BR compared with H.264/AVC.
In human visual system the spatial resolution of a scene under view decreases uniformly at points of increasing distance from the point of gaze, also called foveation point. This phenomenon is referred to as foveation...
详细信息
ISBN:
(纸本)0819456586
In human visual system the spatial resolution of a scene under view decreases uniformly at points of increasing distance from the point of gaze, also called foveation point. This phenomenon is referred to as foveation and has been exploited in foveated imaging to allocate bits in image and video coding according to spatially varying perceived resolution. Several digital image processing techniques have been proposed in the past to realize foveated images and video, In most cases a single foveation point is assumed in a scene[1-3]. Recently there has been a significant interest in dynamic as well as multi-point foveation. The complexity involved in identification of foveation points is however significantly high in the proposed approaches [4-5]. In this paper, an adaptive multi-point foveation technique for video data based on the concepts of regions of interests (ROIs) is proposed and its performance is investigated. The points of interest are assumed to be centroid of moving objects and dynamically determined by the foveation algorithm proposed. Fast algorithm for implementing region based multi-foveation processing is proposed. The proposed adaptive multi-foveation fully integrates with existing video codec standard in both spatial and DCT domain.
In order to efficiently transmit video data while satisfying the channel bandwidth and transmission delay constraints, bit rate control of the video encoding process is required. According to ultra-high-definition vid...
详细信息
ISBN:
(纸本)9789811381386;9789811381379
In order to efficiently transmit video data while satisfying the channel bandwidth and transmission delay constraints, bit rate control of the video encoding process is required. According to ultra-high-definition video, traditional coding algorithms have a large amount of caculation and high computational complexity, thus, parallel coding methods such as inter-frame parallel coding and wavefront parallel coding (WPP) are proposed. However, the rate control of parallel coding is a difficult problem, especially the intra-frame rate control under the WPP coding mode, so this paper proposes a bit rate control algorithm within macro-block layer. By contrasting the PSNR, the encoding speed and the VBV (Video Buffer Verifier) buffer condition of the video sequences, the algorithm proposed in this paper has advantages of less computing cost and faster coding speed than the traditional algorithms.
In the paper is presented one new approach for efficient presentation of video sign language interpretations, used in training of hearing impaired people. The idea is based on the use of contour image sequences instea...
详细信息
ISBN:
(纸本)9788022728560
In the paper is presented one new approach for efficient presentation of video sign language interpretations, used in training of hearing impaired people. The idea is based on the use of contour image sequences instead of the original color ones. The aim is to achieve efficient compression, which to offer easier access for distance learning applications or mobile communications. The contours extraction is based on image filtration and background equalization, followed by image segmentation and lossless intra-frame compression of the consecutive TV frames of the video interpretations. In result, the understandability of the sign language interpretations is retained. The high compression ratio obtained ensures easier accessibility for the presented information. The comparison with other similar methods proved the efficiency of the new approach.
A multiple camera surveillance system is typical example of a Distributed Video (DV) system with many-to-one topology which demands the use of new paradigm with multiple encoders, installed at number of locations and ...
详细信息
A multiple camera surveillance system is typical example of a Distributed Video (DV) system with many-to-one topology which demands the use of new paradigm with multiple encoders, installed at number of locations and very few decoders in a control room. This paradigm open-up new frontiers for the research community in designing low cost encoders even at the cost of expensive decoders. A solution to this problem based on information theory finding of 70s, by Slepian and Wolf [3], for lossless encoding, and followed by the work of Wyner and Ziv [5], for lossy encoding. In the last few years there has been significant research activity in the design and implementation of video codec based on these findings. Our research work is also part of that effort. In this paper we will present a modified Wyner-Ziv codec, that take advantage of slow motion activity, which is typically of surveillance data. GOP is selected dynamically by accessing motion activity and also puncturing bit rate varies adaptively. Also the proposed architecture takes advantage of intra-frame coding for both key frame and Wyner-Ziv frames.
暂无评论