Dissertation > Excellent graduate degree dissertation topics show

The Research of Facial Expression Recognition Based on GPU High Performance Computing

Author: XiaChunFen
Tutor: ZouChengMing
School: Wuhan University of Technology
Course: Computer Science and Technology
Keywords: GPU CUDA High Performance Computing Facial Expression Recognition
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 248
Quote: 2
Read: Download Dissertation

Abstract


In the field of digital image processing, with the research content and the increasing complexity of the algorithm, time and resource consumption is growing, this in-depth study in this field has brought great challenges, the graphics processor (GPU) enable the rapid development it could be outside the graphics process provides a good general-purpose computing platform. NVIDIA Corporation issued Compute Unified Device Architecture (CUDA) can effectively utilize the processing power of GPU strong and huge memory bandwidth for graphics rendering calculations outside, are widely used in various fields of modern science and technology. On the other hand, face recognition is a pattern recognition and image processing is the most cutting-edge research topics, especially facial expression recognition in this area, because facial expressions are very complicated, the computer recognizes it is not easy, so this paper uses a statistical methods for identification, the method of data-intensive, computing capacity, high repeatability, with the typical characteristics of parallel computing, this article presents a GPU-based high-performance platform for facial expression recognition algorithm and optimization methods. This paper analyzes the GPU architecture and CUDA relevant theoretical knowledge, the parallelization of the digital image analysis carried out by the GPU-based algorithm for image binarization experiment proved that the GPU parallel processing of digital images has obvious advantages. The general design of the GPU and CUDA architecture calculation method of facial expression recognition problem for high-performance computing solutions, with 216 stream processors based on the GTX 260's GPU for facial expression recognition method for parallel implementation, the algorithm with the original CPU calculation accuracy of the results under the same conditions, the use of GPU parallel implementation efficiency can be increased 220-fold, the experimental results show that the GPU-based high-performance computing in the processing of facial expression recognition is very effective and can significantly improve computational efficiency. On the GPU shared memory and texture memory block thread synchronization techniques to study relative to the global memory, the texture memory storage capacity, can fully meet the requirements of face data stored in the shared memory GPU chip delay between the threads only global memory 1/100, inter-thread access speed is very fast, full use of local memory resources is to achieve minimal latency inter-thread communication methods. Using this method, a GPU-based optimization algorithm to solve the problem of delay between GPU threads can further improve efficiency, the relative CPU speed nearly 700 times can be calculated, the results show that GPU computing for large-scale data parallel computing has a strong ability to adapt In order to improve the efficiency of pattern recognition provides a new way.

Related Dissertations

  1. Visual Feedback and Memory Behavior Based GPU Parallel Ant Colony Algorithm,TP301.6
  2. Alexithymia individual's facial expression recognition and attentional bias experimental study,B849
  3. Research of Finite Element Method on GPU,O241.82
  4. Research of Image Segmentation in Web Image Search Based on GPUs,TP391.41
  5. GPU-based SIFT algorithm,TP391.41
  6. CUDA-based regular expression matching system design and implementation,TP311.52
  7. GPU-based image search Chinese Research on key technologies of the retrieval,TP391.1
  8. Based on GPU / CPU multi- level parallel CFD Optimization Strategy,V221
  9. White Light Interferometry for Fast Areal Surface Measurement Based on GPGPU,O439
  10. Ffmpeg-based high-performance high-definition streaming media player software design,TN919.8
  11. Flame effects Augmented Reality System Key Technologies,TP391.9
  12. Efficient multi- GPU based acoustic wave simulator and its application,TP391.41
  13. Research groups and evacuation simulation algorithm simulation system,TP391.9
  14. GPU-accelerated particle filter PET image reconstruction algorithm,TP391.41
  15. GPU-based book recommendation system Research and Implementation,TP391.3
  16. GPU-based acceleration of a linear programming algorithm and its application,TP391.41
  17. CUDA-based video fire detection system,TP391.41
  18. GPU-based parallel search algorithm for time series,TP391.41
  19. CPU-based inverse algorithm source strength,TP18
  20. GPU-based reconstruction algorithm for X-rays accelerate research,TP391.41
  21. GPU-based acceleration of neutral gas leak and rescue simulation study,TP391.41

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile