分形维数计算的流水线优化方法研究

Study on the pipeline optimization method for fractal dimension calculation

分形维数计算具有计算复杂度高、计算时间长等特点,严重影响计算的实时性。针对此问题,在充分分析分形维数计算内在特性的基础上,利用分形维数具有流水线计算的特点,提出了一种计算分形维数的流水线体系结构,可有效提高分形维数计算的实时性。由于嵌入式并行处理硬件平台资源有限,对分形维数计算实时性进行优化的同时还需要考虑资源消耗的优化。通过对不同级数流水线运行时间和资源消耗的分析,建立基于运行时间与资源消耗的优化目标函数,从而得到运行时间与资源消耗最优的流水线结构。并与已有的计算分形维数的并行算法进行对比分析,实验结果表明,本文提出的优化方法在提高计算实时性的同时有效降低了资源消耗,实现了运行时间与资源消耗的优化。

Fractal dimension calculation has the characteristics of high calculation complexity and long calculation time,which seriously affects the real time performance of the fractal dimension calculation. Aiming at this problem,a pipeline system structure for fractal dimension calculation is proposed based on sufficiently analyzing the internal characteristic of fractal dimension calculation,using the characteristic that fractal dimension has the characteristic of pipeline calculation. The pipeline system structure can improve the real time performance of fractal dimension calculation. Because of the limited resource of the embedded parallel processing hardware platform,the optimization of the resource consumption is another factor that needs to be considered while the fractal dimension calculation real time performance is optimized. Through analyzing the run time and resource consumption of the pipelines with different stages,the optimized objective function based on run time and resource consumption is established. According to the optimized objective function,a pipeline with the optimal run time and resource consumption is obtained. The proposed method was compared and analyzed with existing parallel algorithms of fractal dimension calculation. Experiment results indicate that the proposed pipeline optimization algorithm for fractal dimension calculation improves the calculation real time performance and decreases the resource consumption effectively at the same time.The optimization of both run time and resource consumption in fractal dimension calculation is achieved.