Dissertation > Excellent graduate degree dissertation topics show

Motion Estimation in H.264 Implementation of Parallel Computing

Author: WanShuang
Tutor: LiuHaiHua
School: Central South University for Nationalities
Course: Biomedical Engineering
Keywords: H.264/AVC Parallel Computing CUDA GPGPU Motion Estimation
CLC: TN919.81
Type: Master's thesis
Year: 2009
Downloads: 218
Quote: 3
Read: Download Dissertation


With the rapid development of information technology, people also will increase demand for information and video information because of its intuitive features which are widely used, there are already many international standard for video compression, including the latest video standard is H .264/AVC video compression standard, which has a higher compression ratio because of the wide range of applications. However, H.264/AVC video coding standard is currently the highest complexity, and its coding structure in addition to inheriting the traditional coding module also adds a lot of new modules, which greatly increases the amount of computation. Therefore, H.264/AVC study focuses on how to speed up encoding speed. The conventional motion estimation in video standards using only 1-2 sizes of macro-blocks, using a variety of sizes H.264/AVC macroblock motion estimation so that the share in the coding operation to further increase the proportion of the amount. Therefore, the need to adopt effective structure or method to achieve motion estimation algorithm, thus speeding up the encoding rate. With the GPU (Graphic Processing Unit, graphics processor) the rapid development, more and more parallel processing-intensive operations can be ported to the GPU. In this paper, the platform, parallel processing for motion estimation, and its innovations include the following aspects: (1) gives the full search motion estimation algorithm for parallel implementation. GPU multithreaded architecture according to the characteristics of each macro block motion estimation in 256 threads running, obtained 1024 8 × 8 macroblock SAD values, and these SAD values ​​into 16 × 16, etc. Macro block SAD value, and finally parallel search, obtain the minimum SAD value. Experimental results show that the parallel motion estimation based on GPU can more effectively improve encoding speed. (2) gives the integer DCT (Discrete Cosine Transform) transform of the parallel algorithm. The algorithm still use multithreading GPU features twice the residual matrix parallel matrix multiplication, and parallel matrix multiplication is the key to the use of 4 × 4 threads compute the corresponding transform coefficients, which uses only addition, subtraction and shift operation. Experiments show that the parallel integer DCT transform algorithms based on the CPU integer DCT transform algorithm more advantages. To further verify the GPU-based parallel coding performance, this paper built based on open-source encoder X264 experimental platform, first in the X264 encoder configuration code can be compiled CUDA runtime environment, and then modify the X264 in the serial data processing mode change GPU processing suitable for parallel processing mode, and then use the proposed parallel algorithm for motion estimation X264 rewrite the relevant code, and finally using the generated executable file for the video sequences tested. The results showed that the GPU parallel coding on the CPU operation speed is 1.5 times, and improved image quality. To sum up: By H.264/AVC video standard and GPU parallel processing in-depth analysis, in order to improve the speed of encoding is proposed based on CUDA platform parallel motion estimation algorithm and parallel integer DCT transform algorithm to solve the H.264 / AVC video coding standard high-resolution images in the time-consuming part of the problem. Parallel experiments show that the proposed motion estimation algorithm and parallel integer transform algorithm to achieve the desired design requirements, H.264 systems constructed to achieve a certain degree of real-time video encoding.

Related Dissertations

  1. Rate-distortion Optimization Based Rate Control,TN919.81
  2. Optimizing and Realising Research on Vedio Compression in TV Guidance System,TN919.81
  3. Research on Rotational Image Stabilization Algorithms for Electronic Image Stabilization System,TP391.41
  4. Study on Side Information Generation in Distributed Video Coding,TN919.81
  5. Research and Design of a High-Performance Scalable Public Key Cryptographic Coprocessor,TN918.1
  6. Research on Video Compression Algorithm Based on Multi-core Computing Platform,TN919.81
  7. Research on Mode Decision Algorithm for H.264/AVC Intra Prediction,TN919.81
  8. Research on Parallel Computation of H.264 Based on CUDA,TN919.81
  9. Study of Adaptive Motion Estimation in H.264/AVC,TN919.81
  10. Study of Binocular Stereo Matching Algorithm and Algorithm Implementation Based on Multi-core.,TP391.41
  11. Spatial Uncertainty in Stereo Visual Odometry,TP391.41
  12. Research of Finite Element Method on GPU,O241.82
  13. Numerical Simulation of Radiofrequency Waves in Magnetized Plasma,TL612
  14. The Algorithm Researches of Novel Wide Area Backup Protection for Power Grid,TM774
  15. The Research on Online Adaptive Settings,TM77
  16. H.264 Motion Estimation Algorithm and Application of hierarchical coding,TN919.81
  17. CUDA-based regular expression matching system design and implementation,TP311.52
  18. Fault Tolerance for MapReduce in the Cloud Environment,TP302.8
  19. Implement and Research of High-Quality Global Illumination on GPU,TP391.41
  20. Study on Real-time Screen Space Fluid Rendering,TP391.41
  21. White Light Interferometry for Fast Areal Surface Measurement Based on GPGPU,O439

CLC: > Industrial Technology > Radio electronics, telecommunications technology > Communicate > Image communication, multimedia communication > Image coding
© 2012 www.DissertationTopic.Net  Mobile