A5046233912- Video Coding and Compression Technologies
- Advanced Data Compression Techniques
- Advanced Image Processing Techniques
- Advanced Vision and Imaging
- Image and Signal Denoising Methods
- Image and Video Quality Assessment
- Image Enhancement Techniques
- Advanced Image and Video Retrieval Techniques
- Video Analysis and Summarization
- Medical Image Segmentation Techniques
- Image Retrieval and Classification Techniques
- Digital Filter Design and Implementation
- Advanced Wireless Network Optimization
- Embedded Systems Design Techniques
- Visual Attention and Saliency Detection
- Advanced Optical Sensing Technologies
- Telecommunications and Broadcasting Technologies
- CCD and CMOS Imaging Sensors
- Sparse and Compressive Sensing Techniques
- Sensor Technology and Measurement Systems
- Neural Networks and Applications
- Advanced Bandit Algorithms Research
- Color Science and Applications
- Video Surveillance and Tracking Methods
- Industrial Vision Systems and Defect Detection
Google (United States)
2015-2024
Jiangxi University of Finance and Economics
2023
University of California, Santa Barbara
2010-2014
Gwangju Institute of Science and Technology
2013
Tsinghua University
2013
AV1 is an emerging open-source and royalty-free video compression format, which jointly developed finalized in early 2018 by the Alliance for Open Media (AOMedia) industry consortium. The main goal of development to achieve substantial gain over state-of-the-art codecs while maintaining practical decoding complexity hardware feasibility. This paper provides a brief technical overview key coding techniques along with preliminary performance comparison against VP9 HEVC.
Google has recently finalized a next generation open-source video codec called VP9, as part of the libvpx repository WebM project (http://www.webmproject.org/). Starting from VP8 released by in 2010 baseline, various enhancements and new tools were added, resulting next-generation VP9 bit-stream. This paper provides brief technical overview along with comparisons other state-of-the-art codecs H.264/AVC HEVC on standard test sets. Results show to be quite competitive mainstream codecs.
The AV1 video compression format is developed by the Alliance for Open Media consortium. It achieves more than a 30% reduction in bit rate compared to its predecessor VP9 same decoded quality. This article provides technical overview of codec design that enables performance gains with considerations hardware feasibility.
Google has recently finalized a next-generation open-source video codec called VP9, as part of the libvpx repository WebM project (http://www.webmproject.org/). Starting from VP8 released by in 2010 baseline, various enhancements and new tools were added, resulting bit stream VP9. The was with exception essential bug fixes June 2013. Prior to release, however, all technical developments being conducted openly public experimental branch for many months. This paper provides brief overview...
Deep learning has revolutionized many computer vision fields in the last few years, including learning-based image compression. In this paper, we propose a deep semantic segmentation-based layered compression (DSSLIC) framework which segmentation map of input is obtained and encoded as base layer bit-stream. A compact representation also generated first enhancement layer. The version are then employed to obtain coarse reconstruction image. residual between additionally another Experimental...
Recently deep learning-based image compression methods have achieved significant achievements and gradually outperformed traditional approaches including the latest standard Versatile Video Coding (VVC) in both PSNR MS-SSIM metrics. Two key components of learned are entropy model latent representations encoding/decoding network architectures. Various models been proposed, such as autoregressive, softmax, logistic mixture, Gaussian Laplacian. Existing schemes only use one these models....
This paper proposes a novel approach to jointly optimize spatial prediction and the choice of subsequent transform in video image compression. Under assumption separable first-order Gauss-Markov model for signal, it is shown that optimal Karhunen-Loeve Transform, given available partial boundary information, well approximated by close relative discrete sine (DST), with basis vectors tend vanish at known maximize energy unknown boundary. The overall intraframe coding scheme thus switches...
In 2018, the Alliance for Open Media (AOMedia) finalized its first video compression format AV1, which is jointly developed by industry consortium of leading technology companies.The main goal AV1 to provide an open source and royalty-free coding that substantially outperforms state-of-the-art codecs available on market in efficiency while remaining practical decoding complexity as well being optimized hardware feasibility scalability modern devices.To give detailed insights into how...
Recently, deep learning-based image compression has made significant progresses, and achieved better rate-distortion (R-D) performance than the latest traditional method, H.266/VVC, in both MS-SSIM metric more challenging PSNR metric. However, a major problem is that complexities of many leading learned schemes are too high. In this paper, we propose an efficient effective coding framework, which achieves similar R-D with lower complexity state art. First, develop improved multi-scale...
This paper proposes a new approach to combined spatial (Intra) prediction and adaptive transform coding in block-based video image compression. Context-adaptive from available, previously decoded boundaries of the block, is followed by optimal residual. The derivation both for error, assumes separable first-order Gauss-Markov model signal. resulting shown be close relative sine with phase frequencies such that basis vectors tend vanish at known maximize energy unknown boundaries. overall...
Google has recently been developing a next generation opensource video codec called VP9, as part of the experimental branch libvpx repository included in WebM project (http://www.webmproject.org/). Starting from VP8 released by 2010 baseline, number enhancements and new tools have added to improve coding efficiency. This paper provides technical overview current status this along with comparisons other stateoftheart codecs H. 264/AVC HEVC. The that so far include: larger prediction block...
Google has recently finalized a next generation open-source video codec called VP9, as part of the libvpx repository WebM project (http://www.webmproject.org/). Starting from VP8 released by in 2010 baseline, various enhancements and new tools were added, resulting next-generation bit-stream VP9. The was with exception essential bug-fixes, June 2013. Prior to release however, all technical developments fact being conducted openly public experimental branch for many months. This paper...
The quality fluctuation of video is significant in human visual system, and thus, many rate control schemes are widely developed the area communication. In recent years, researchers show more interests region interest (ROI)-based encoding, it applied latest codecs, such as HEVC VP9. This paper presents a new scheme for ROI mode coding based on discrete fourier transform coefficient model radial basis function neuron network. A R-D proposed by classifying blocks into different depth, groups,...
Learned image compression has recently shown the potential to outperform standard codecs. State-of-the-art rate-distortion (R-D) performance been achieved by context-adaptive entropy coding approaches in which hyperprior and autoregressive models are jointly utilized effectively capture spatial dependencies latent representations. However, latents feature maps of same resolution previous works, contain some redundancies that affect R-D performance. In this paper, we propose first learned...
Recently deep learning-based image compression has shown the potential to outperform traditional codecs. However, most existing methods train multiple networks for bit rates, which increases implementation complexity. In this paper, we propose a variable-rate framework, employs more Generalized Divisive Normalization (GDN) layers than previous GDN-based methods. Novel residual sub-networks are also developed in encoder and decoder networks. Our scheme uses stochastic rounding-based scalar...
A novel intra prediction scheme, based on recursive extrapolation filters, is introduced. Standard largely consists of copying boundary pixels (or linear combinations thereof) along certain directions, which reflects an overly simplistic model for the underlying spatial correlations. As alternative, we view image signal as a 2-D non-separable Markov model, whose corresponding correlation better captures nuanced directionality effects within blocks. This viewpoint motivates design set modes...
Google started an opensource project, entitled the WebM Project, in 2010 to develop royaltyfree video codecs for web. The present generation codec developed project called VP9 was finalized mid2013 and is currently being served extensively by YouTube, resulting billions of views per day. Even though adoption outside still its infancy, has already embarked on ambitious a next edition VP10 that achieves at least generational bitrate reduction over current VP9. Although early stages, set new...
Recently deep learning-based image compression has shown the potential to outperform traditional codecs. However, most existing methods train multiple networks for bit rates, which increase implementation complexity. In this paper, we propose a new variable-rate framework, employs generalized octave convolutions (GoConv) and transposed-convolutions (GoTConv) with built-in divisive normalization (GDN) inverse GDN (IGDN) layers. Novel GoConv- GoTConv-based residual blocks are also developed in...
Contemporary lossy image and video coding standards rely on transform coding, the process through which pixels are mapped to an alternative representation facilitate efficient data compression. Despite impressive performance of end-to-end optimized compression with deep neural networks, high computational space demands these models has prevented them from superseding relatively simple found in conventional codecs. In this study, we propose learned transforms entropy that may either serve as...
Conventional "pixel copying" prediction used in current video standards was shown previous work to be sub-optimal compared 2-D non-separable Markov model based recursive extrapolation approaches. The premise of this paper is that order achieve the full potential these approaches it necessary account for several requirements, namely, design modes (and respective filters) must optimize a rate-distortion cost rather than minimize mean squared error; filters sufficient complexity cover all...
Video Multi-method Assessment Fusion (VMAF) is a machine-learning based video quality metric. It experimentally shown to provide higher correlation with human visual system as compared conventional metrics like peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) in many scenarios has drawn considerable interest an alternative metric evaluate the perceptual quality. This work proposes systematic approach improve compression performance VMAF. composed of multiple...
Recently, vision transformers have been applied in many computer problems due to its long-range learning ability. However, it has not throughly explored image compression. We propose a patch-based learned compression network by incorporating transformers. The input is divided into patches before feeding the encoder and are reconstructed from decoder form complete image. Different kinds of transformer blocks (TransBlocks) meet various requirements subnetworks. also transformer-based context...
Recently deep learning-based image compression methods have achieved significant achievements and gradually outperformed traditional approaches including the latest standard Versatile Video Coding (VVC) in both PSNR MS-SSIM metrics. Two key components of learned are entropy model latent representations encoding/decoding network architectures. Various models been proposed, such as autoregressive, softmax, logistic mixture, Gaussian Laplacian. Existing schemes only use one these models....
Learned image compression has recently shown the potential to outperform standard codecs. State-of-the-art rate-distortion (R-D) performance been achieved by context-adaptive entropy coding approaches in which hyperprior and autoregressive models are jointly utilized effectively capture spatial dependencies latent representations. However, latents feature maps of same resolution previous works, contain some redundancies that affect R-D performance. In this paper, we propose a learned...