电子器件2026，Vol.49Issue(2)：299-304,6.DOI:10.3969/j.issn.1005-9490.2026.02.010
基于HOG特征提取的车辆检测系统的FPGA实现
FPGA Implementation of HOG Feature Extraction for Vehicle Detection
摘要
Abstract
Developing vehicle detection controllers in embedded systems is a complex task due to limited hardware resources.The Histo-gram of orientation gradients has been widely used to detect and match features.However,the HOG algorithm is computationally deman-ding and its implementation in embedded systems requires more efficient methods.An optimised pipeline architecture for HOG feature ex-traction is designed and the system is deployed on an FPGA for vehicle detection.Firstly,a hardware acceleration module for HOG feature extraction is designed on the FPGA side of Zynq.Secondly,a support vector machine module is designed on the ARM side for feature clas-sification.Finally,the AXI4 bus is used to connect the FPGA and ARM for data interaction between hardware and software.Experimental results show that the system accelerates the recognition process while consuming less hardware resources than similar existing work.On the Xilinx Zynq-7020 platform,the system proposed is able to reach a processing speed of 64 FPS using an 800×480 pixel image resolution,while the classification accuracy is 91.86%,which is of good practical value in vehicle control applications.关键词
车辆检测/HOG特征/FPGA/目标识别/硬件设计Key words
vehicle detection/HOG feature/FPGA/target identification/hardware implementation分类
信息技术与安全科学引用本文复制引用
庞宇,杨家斌,王慧倩,张洋..基于HOG特征提取的车辆检测系统的FPGA实现[J].电子器件,2026,49(2):299-304,6.基金项目
重庆市教委科学技术研究项目(KJQN202100602) （KJQN202100602）
中国博士后科学基金资助项目(2022MD713702) （2022MD713702）
