| Hausdorff distance is an important point characteristics matching method. Widely used in image processing, astronomy, mathematics, network applications, medical, economic forecasts, and other fields, especially have a very wide range of applications in the image matching. The traditional Hausdorff distance algorithm has complex calculation and low efficiency, improves the efficiency of the algorithm is useful and important. On the other hand, multi-core technology is the current hot researching field of computer, and significantly improves the computer's calculation ability. It will be a widely popular computing model. However, to really highlight the advantages of multi-core processors, software development must keep up with the pace of hardware, how to develop in line with the multi-core software is increasingly becoming a hot research field of computer technology.This paper first studies the current computer hardware and software technology development trends, then studies the paralleled way around the Hausdorff distance algorithm, aiming at finding a simple and efficient parallel support scheme on multi-core platform. This paper first introduces multi-core technology, multi-core architecture, common developing environment and application of software tools for comparative analysis, then finds the technologies and solutions suitable for play on the performance of multi-core programming. After a detailed analysis of the Hausdorff distance algorithm, designs a Multi-core structure of parallel algorithms, and successfully applied to the building matching and locating system, and also can apply to the printing plate detection system; Next use OpenMP shared memory programming, the combination of Intel VTune Performance Analyzer, Intel Thread Checker and Intel C++ Compiler, and other tools and solutions for developing and testing, then according to the performance in the framework of multi-core platform to make corresponding adjustments, and make two aspects of optimization from the code Parallel and compiler optimization; Finally, according to Amdahl's and Gustafson's Law to make expansion analysis, and make an objective evaluation. One characteristic innovate is making the Hausdorff distance algorithm from the traditional serial method to multi-core parallel computing way suitable for the multi-core structure of the IA's high-performance structure, and successfully applied to the buildings matching and locating system and the printing plate detection system, the other is the performance analysis of parallel code by the newest technology, which realizes algorithm optimization, code level optimization and compiler level optimization, enhances the operation efficiency and the adaptation of hardware development. The technical route used in the subject has the universality, may promotes to the parallelization of other function libraries. |
PDF Full Text Request