基于HOG和SVM行人检测算法的FPGA实现
FPGA Implementation of pedestrian detection based on HOG and SVM
行人检测技术目前已广泛应用于人们的生活和工作中,在不同的场合对行人检测技术提出不同的要求。由于在某些特定场合对行人检测的实时性和功耗有很高要求,本文提出了一种基于HOG (Histogram of oriented gradients)和SVM (Support Vector Machine)的行人检测的FPGA实现。在设计中采用了查找表、流水线技术等方法加速。同时,由于各个步骤的时间消耗的不平衡,采用了模块复用的方法。通过上述的方法,运算速度有了明显的提升。
Nowadays pedestrian detection technology is widely employed in people's daily life and work , however different application scenarios put forward different requirements . A FPGA Implementation of pedestrian detection based on HOG and SVM is proposed in order to achieve a high requirement of instantaneity and power dissipation in some specific occasions . Some accelerated approaches , look up table and pipeline , are adopted . Meanwhile , due to the imbalance of time consumption at various steps , the module reuse method was adopted . The computing speed has the obvious improvement , integrated the method above .
王振宇、朱秋龙
计算技术、计算机技术电子技术应用微电子学、集成电路
行人检测HOGSVMFPGA
pedestrian detectionHOGSVMFPGA
王振宇,朱秋龙.基于HOG和SVM行人检测算法的FPGA实现[EB/OL].(2014-11-04)[2025-08-05].http://www.paper.edu.cn/releasepaper/content/201411-49.点此复制
评论