基于随机计算的边缘检测系统的设计与实现OA
Design and implementation of edge detection system based on stochastic computing
随机计算(stochastic computing,SC)能够通过简单的逻辑实现复杂的运算,其通过牺牲一定的精度,换取了低功耗、低硬件开销以及高容错的优势.本文提出了一种随机计算的边缘检测电路,通过简化随机计算的Sobel算子和截断随机数源等方法,在保证检测质量的前提下有效地降低资源消耗.实验结果表明,本文的研究相比于精确计算,在检测结果无明显差别的前提下,边缘检测部分查找表(look-up table,LUT)降低约54.4％,触发器(flip-flop,FF)降低约 40.4％,数字信号处理单元(digital signal processing,DSP)从 6个降到了 1个.
Stochastic computing(SC)can achieve complex operations through simple logic,sacrificing a certain degree of accuracy in exchange for the advantages of low power consumption,low hardware overhead,and high error toler-ance.This paper proposes a stochastic computing edge detection circuit that effectively reduces resource consump-tion while ensuring detection quality by simplifying the Sobel operator of stochastic computing and truncating random number source.The experimental results show that compared with accurate computation,this study reduces the look-up table(LUT)of the edge detection part by approximately 54.4％,the flip-flop(FF)by approximately 40.4％,and the digital signal processing(DSP)from 6 to 1 without significant differences in detection results.
朱昌瑞;王韬;班恬
南京理工大学电子工程与光电技术学院 南京 210094南京理工大学电子工程与光电技术学院 南京 210094南京理工大学电子工程与光电技术学院 南京 210094
随机计算边缘检测现场可编程门阵列低差异序列图像处理
stochastic computingedge detectionfield programmable gate arraylow discrepancy sequenceimage processing
《高技术通讯》 2026 (1)
80-88,9
江苏省自然科学基金(BK20221486)资助项目.
评论