江西科学2025，Vol.43Issue(4)：680-687,8.DOI:10.13990/j.issn1001-3679.2025.04.015
基于改进两帧差分算法的运动车辆检测
Moving Vehicle Detection Based on An Improved Two-Frame Difference Algorithm
摘要
Abstract
To address the issue of internal voids in detected targets when using the traditional two-frame difference algorithm for moving vehicle detection,this paper proposes an im-proved two-frame difference vehicle detection algorithm incorporating a clustering method.First,clustering is performed on the result of the two-frame difference operation based on the density of foreground pixels,and the clustering results are refined by assigning rectan gular bounding boxes to each cluster.Then,a vehicle edge extraction method is designed by combining the K-means and Canny algorithms,which extracts the target contour inside each rectangular box to complete foreground detection.The experimental results show that the improved algorithm achieves an average precision rate of 0.901,an average recall rate of 0.872,and an average F1 value of 0.886 for detection in two different scenes.Compared with the traditional two-frame difference method,the improved algorithm has an average improvement of 0.325 in these three metrics,significantly enhancing detection performance and effectively addressing the issue of internal voids.关键词
运动车辆检测/两帧差分/聚类算法/边缘提取/空洞Key words
moving vehicle detection/two-frame difference/clustering algorithm/edge ex-traction/voids分类
信息技术与安全科学引用本文复制引用
舒兆翰,程鸿超,潘一平,周烨,朱小飞..基于改进两帧差分算法的运动车辆检测[J].江西科学,2025,43(4):680-687,8.基金项目
浙江省自然资源厅自然资源科技项目(2024ZJDZ020) （2024ZJDZ020）
湖州市科技攻关计划项目(2023GS02). （2023GS02）
