Copyright © 2003-2012 ATC Labs. All Rights Reserved.
Search ATC Labs
Licensing / Support
For licensing details please contact us at sales@atc-labs.com
For details on implementations on other platforms please contact us at info@atc-labs.com
The recently developed video coding standard, H.264 / AVC, significantly outperforms previous standards in terms of coding performance at reasonable implementation complexity. Several application systems, such as high-definition DVD and digital video broadcasting for handheld devices and high-definition television systems, have adopted H.264 or its modified versions as the video coding standard. In addition, the extensions of H.264 / AVC to scalable and multi-view video coding applications are nearly finalized. Many video services, especially bandwidth-limited wireless video, will benefit from the H.264 coder due to its outstanding features. The use of variable block sizes for intra- and inter-prediction in combination with different prediction modes and motion compensation using multiple reference frames is one of the main reasons for the improved coding efficiency in H.264 / AVC. Together with many other new features, the encoder can select between multitudes of different coding modes. The determination of the optimal coding mode under the joint rate and distortion consideration, which is called the rate-distortion optimization (RDO), introduces a huge amount of memory access and computational complexity for testing all possible modes in video encoders. Hence, a reduction of the complexity of motion estimation and mode selection in an H.264/AVC encoder becomes an important task for real-time applications.
In selecting the quantization parameters at the frame and the block levels, the goal is to design a rate-control method that maximizes the video quality with a constrained bandwidth. Regardless of the superior coding efficiency of the H.264 / AVC standard, there still exist many video coding standards. For example, MPEG-2 and H.263 have been adopted by the current television and video telephony systems, respectively. Therefore, an effective transcoding method, which can effectively convert the existing non-H.264 bitstreams to H.264 conforming bitstreams, while maintaining the excellent rate-distortion performance, will greatly smooth the transition in the migration to H.264 / AVC. Therefore our technology focus covers a wide spectrum including the following important topics :
Optimal hierarchical mode decision framework for efficient intra/inter prediction
Fast motion estimation
Fast rate-distortion optimization (RDO)
Rate control based on content analysis
Intra(bit-rate, Resolution, PSNR) and Inter-format(MPEG2/H.263+) transcoding
Quality improvement through efficient perceptual video measures
Fast entropy coding/decoding
In selecting the quantization parameters at the frame and the block levels, the goal is to design a rate-control method that maximizes the video quality with a constrained bandwidth. Regardless of the superior coding efficiency of the H.264 / AVC standard, there still exist many video coding standards. For example, MPEG-2 and H.263 have been adopted by the current television and video telephony systems, respectively. Therefore, an effective transcoding method, which can effectively convert the existing non-H.264 bitstreams to H.264 conforming bitstreams, while maintaining the excellent rate-distortion performance, will greatly smooth the transition in the migration to H.264 / AVC. Therefore our technology focus covers a wide spectrum including the following important topics :
Optimal hierarchical mode decision framework for efficient intra/inter prediction Fast motion estimation Fast rate-distortion optimization (RDO) Rate control based on content analysis Intra(bit-rate, Resolution, PSNR) and Inter-format(MPEG2/H.263+) transcoding Quality improvement through efficient perceptual video measures Fast entropy coding/decoding
A novel partial prediction approach for fast Intra-prediction mode decision based on pixel correlation and inherent symmetry which exists in different prediction modes brings down the number of prediction modes from 9 to 3 for robust rate-distortion evaluation in case of 4x4 blocks and thus reducing complexity by almost 65%.
Jointly optimized inter-mode selection and motion estimation significantly reduce the encoder complexity.
Statistical estimations of actual bit consumption and the reconstruction distortion from spatial and transform domain features achieve similar performance to robust Lagrange rate distortion optimization with exhaustive full search while considerably reducing the computation time.
A codec independent rate-control framework helps to provide optimum playback settings or constant quality encoding (Variable bit-rate, VBR) or meet total bit-budget constraints in case of encoding for storage applications to DVD etc. It also helps to meet tight constraints of buffer sizes, decoding delays and provide constant bit-rate (CBR) support (for a specified bit-rate) for streaming or real-time multimedia distribution services. It can be one pass or two pass (first one for complete content analysis) solution depending on the end requirements like high quality or real-time support etc.
An optimized common encoding/decoding core offers extra flexibility for intra and inter-format transcoding.
Efficient probability based techniques for optimal encoding/decoding of codewords exploiting the symmetry in codeword lengths significantly speed up the process of memory accesses in look-up tables.
Though basic distortion measures like mean-squared error (MSE) or peak signal-to-noise ratio (PSNR) are simple and very popular, they do not correlate well with perceived quality. Low complexity perceptual quality based measures offer optimal quality for digital imaging systems with respect to the capture, display, storage and transmission of visual information.
Jointly optimized inter-mode selection and motion estimation significantly reduce the encoder complexity.
Statistical estimations of actual bit consumption and the reconstruction distortion from spatial and transform domain features achieve similar performance to robust Lagrange rate distortion optimization with exhaustive full search while considerably reducing the computation time.
A codec independent rate-control framework helps to provide optimum playback settings or constant quality encoding (Variable bit-rate, VBR) or meet total bit-budget constraints in case of encoding for storage applications to DVD etc. It also helps to meet tight constraints of buffer sizes, decoding delays and provide constant bit-rate (CBR) support (for a specified bit-rate) for streaming or real-time multimedia distribution services. It can be one pass or two pass (first one for complete content analysis) solution depending on the end requirements like high quality or real-time support etc.
An optimized common encoding/decoding core offers extra flexibility for intra and inter-format transcoding.
Efficient probability based techniques for optimal encoding/decoding of codewords exploiting the symmetry in codeword lengths significantly speed up the process of memory accesses in look-up tables.
Though basic distortion measures like mean-squared error (MSE) or peak signal-to-noise ratio (PSNR) are simple and very popular, they do not correlate well with perceived quality. Low complexity perceptual quality based measures offer optimal quality for digital imaging systems with respect to the capture, display, storage and transmission of visual information.
Publications :
Y N Sairam, Nan Ma, Neelu Sinha, “A Novel Partial Prediction Algorithm for Fast 4x4 Intra Prediction Mode Decision in H.264”, IEEE Data Compression Conference (DCC), March 25-27, 2008, Utah, USA...Pdf
Y N Sairam, Nan Ma, Neelu Sinha, “A Novel Partial Prediction Algorithm for Fast 4x4 Intra Prediction Mode Decision in H.264”, IEEE Data Compression Conference (DCC), March 25-27, 2008, Utah, USA...Pdf
Technology Highlights :
