光电工程, 2012, 39 (5): 116, 网络出版: 2012-05-31
实时 DSP图像处理空间低通滤波模块优化
The Optimization of the Space Low-pass Filtering Module in the Real-time Image Processing Based on DSP
实时图像处理 程序优化 数字信号处理器 空间低通滤波 real-time image processing the optimization of program DSP space low-pass filtering
摘要
针对实时图像处理系统中预处理核心模块空间低通滤波算法费时的问题,利用 C6416(TMS320C6416)处理器的内部资源,提出了基于该处理器的空间低通滤波程序的设计方案。通过将滤波器进行分解、利用相邻窗口的相关性、浮点除法转化为定点乘法运算等手段,充分复用了中间计算结果,减少了计算与存取数据的次数,发挥了 C6416的指令级并行处理能力,解决了空间低通滤波模块耗时过长的难题。结果证明,对于 256×256大小的处理区域,用时从优化前的 6.0 ms减少到优化后的 0.8 ms,大大缩短了处理时间。本文为 DSP上的实时程序优化提供了重要的经验,给空间滤波类的模块优化提供了一种通用的方法。
Abstract
In the real-time image processing system, in order to solve real-time performance of the space low-pass filtering algorithm, an optimized method based on TMS320C6416 (C6416) was presented. The design applied the technologies like decomposing the low-pass filter, using the relativity of adjacent windows, using the fixed-point multiplication to implement floating-point division, reusing the intermediate result sufficiently and reducing the calculation and access of memory. The optimized algorithm is suitable for realizing on C6416 and solving real-timeperformance of the space low-pass filtering algorithm. The result shows that C6416 spares 0.8 ms after optimization and spares 6 ms before optimization. This research makes significant contributions to the method for optimization of program based on DSP.
雷涛, 曹晓伟, 吴钦章. 实时 DSP图像处理空间低通滤波模块优化[J]. 光电工程, 2012, 39(5): 116. LEI Tao, CAO Xiao-wei, WU Qin-zhang. The Optimization of the Space Low-pass Filtering Module in the Real-time Image Processing Based on DSP[J]. Opto-Electronic Engineering, 2012, 39(5): 116.