A5000179650- Video Coding and Compression Technologies
- Advanced Data Compression Techniques
- Advanced Vision and Imaging
- Advanced Image Processing Techniques
- Advanced Image and Video Retrieval Techniques
- Advanced Steganography and Watermarking Techniques
- Advanced Wireless Communication Techniques
- Image Retrieval and Classification Techniques
- Video Surveillance and Tracking Methods
- Corporate Finance and Governance
- Auditing, Earnings Management, Governance
Tsinghua University
2012-2021
Deakin University
2020
National Engineering Research Center for Information Technology in Agriculture
2019
A concept of a quadtree based nonsquare block coding structure is presented in this paper for the emerging High Efficiency Video Coding standard, which includes asymmetric motion partitioning scheme and transform (NSQT) algorithm. Nonsquare inter frame provides possibility getting more accurate prediction results by splitting one into two blocks. Contrary to traditional symmetric partitions (SMP), (AMP) are proposed improve efficiency, especially blocks with irregular object boundaries. NSQT...
This paper proposes a new intra coding scheme, known as short distance prediction (SDIP), for high efficiency video (HEVC) standardization work. The proposed method is based on the quadtree unit structure of HEVC. By splitting into nonsquare units and reconstruction, therefore shortening distances between predicted reference samples, accuracy can be improved when applying directional method. SDIP improves accuracy, especially high-detailed regions. approach applied in both luma chroma...
This letter focuses on the compression of deep features. With rapid expansion feature data in various CNN-based analysis and processing tasks, demand for efficient continues to increase. Product quantization (PQ) is widely used compact expression In process, vectors are mapped into fixed-length codes based a pre-trained codebook. However, PQ not specifically designed compression, suitable further such as entropy coding. this letter, we propose an entropy-optimized scheme By introducing loss...
This paper proposes a new intra coding scheme, short distance prediction(SDIP), for the current HEVC standardization work. The proposed method is based on quadtree unit structure of HEVC. By splitting unit(CU) into non-square prediction units(PUs) and reconstruction therefore shorten distances between predicted reference pixels, accuracy can be improved when applying directional method. SDIP improves especially high-detailed regions. When integrated software, it shows up to 12.6% bit rate...
Video coding standards commonly use discrete cosine transform (DCT) to the motion compensation (M-C) residuals. However, MC residuals have much weaker correlation than image pixels, and DCT is not optimized for them. In this paper, we propose an adaptive structure Unlike traditional approaches which a predefined core, apply singular value decomposition (SVD) on prediction block eigenvector matrices as core. Experiments show that more efficient compared with approach. An average 2.0% bit rate...
A new non-square quadtree transform structure is proposed to improve the coding efficiency of HEVC video algorithm. The method, designed for inter prediction blocks, incorporates transforms with residual (RQT) structure. It improves adaptation complex image contents through exploiting directional features and contours, breaking limitation RQT when meeting scenarios. Through test, both satisfied subjective objective performance improvement can be achieved there nearly no implementation cost...
In the exploration of next video coding standard, quadtree plus binary tree block structure (QTBT) has been adopted in joint team. QTBT splits a into sub-blocks following or tree. However, all split modes are symmetrical, which is less efficient asymmetrical motion. addition, lower signaling efficiency side information occurs when successive same direction selected. Motivated by these observations, derived mode (DM) based on proposed this letter, where six novel DMs introduced. The DM...
This paper proposes an efficient probability estimation design based on the Logarithmic Binary Arithmetic Coding (LBAC) in second generation of Audio Video coding Standard (AVS2). The process LBAC is not stable enough case violent fluctuation and lack flexibility updating speed. proposed introduces two regulatory factors to mitigate adjust all along line. It can achieve a higher-precision more with hardware-friendly architecture. Experiments are carried out JVET AVS2 reference softwares,...
AVS3 is the next generation video coding standard and being developed by audio (AVS) working group of China. As successor AVS2, designed to achieve a great efficiency beyond AVS2 as well H.265/HEVC. In this paper, we propose novel block partition method for intra coding, named derived tree (DTBP), based on structure in AVS3. By further splitting leaf node quad split types or asymmetric types, signaling flexibility partitioning texture can be improved. When implemented HPM3.1, up 1.14% an...
In the state of art hybrid video coding structure, intra prediction and transform are most important two parts for picture coding. Generally, process separated. The current block is first predicted from neighboring reconstructed pixels, residue then transformed. our approach, combined, therefore more information can be used accurate achieved.