The Parallel Implement Of Douglas-Peucker Curve Compression Algorithm Based On CUDA

Areas such as digital map, positioning and navigation bring a lot of curve data, and these data is so large that it occupys too much storage space and leads to low transmission speed. Concerning the problem, Douglas-Peucker curve compression algorithm becomes a hot research topic.Douglas-Peucker is a classical curve compression algorithm, whose performace is so obvious that it is widely used in many fields. However, the present researches alout the algorithm are mainly limited to serial solutions on CPU. With the data scale increasing, the running time of searial algorithm is too long to meet the requirements. Morever, the effect of serial optimization scheme is not obvious. In this case, there is no doubt that parallelization is a good solution. In 2007, NVIDIA realsed CUDA platform, which plays powerful float point data computing ability of GPU and provides a convenient and easily programmable choice for parallel computing.Based on CUDA, This paper focuses on the Douglas-Peucker algorithm. To understand the domestic and foreign research situation, a lot of realted documents are read and parallel methods for the algorithm are designed. Meanwhile, the parallel schemes of related algorithm are provided, as well as some common optimization schemes summarized.In the end, various types of experiments are made. The results of experiments show that, the algorithm based on CUDA has good effect. Comparing to serial Douglas-Peucker algorithm, this paper presents that parallel algorithm has obvious advantages on the serial one, with the number of points increasing. Therefore, the experiment has engineering and scientific research values.