基于概率统计局部聚类的复杂场景弱小运动目标检测  

Detection of Weak and Small Moving Targets in Complex Scenes Based on Probabilistic Local Clustering

作　　者：何洋 HE Yang(Public Teaching Department,Chuzhou City Vocation College,Chuzhou Anhui 239000,China)

机构地区：[1]滁州城市职业学院公共教学部,安徽滁州239000

出　　处：《菏泽学院学报》2023年第2期12-16,共5页Journal of Heze University

基　　金：安徽省职业与成人教育学会2020年度教育教学研究规划一般课题(azcg162)。

摘　　要：在复杂场景下进行弱小运动目标的检测过程中,受到视频中有序相邻对象相似性的影响,难以高精度地识别低质量特征的运动目标.设计一种基于概率统计局部聚类的复杂场景弱小运动目标检测方法.建立空域模型、时域模型,分析小目标特征、背景特征以及噪声特征,从中提取弱小运动目标特征,使用概率统计局部聚类理论,拟定划分标准使用条件,实现目标轮廓曲线分段,将弱小运动目标检测流程进行整体优化,完成基于概率统计局部聚类的复杂场景弱小运动目标检测方法的设计.检测方法性能测试结果表明:在不同的实验图像下,设计的弱小运动目标检测方法的检测率平均值为69.8%,具有一定的实用性和可靠性.Affected by the similarity of orderly adjacent objects in the video,it is difficult to identify low-quality objects with high accuracy by detecting weak and small moving objects in complex scenes.This paper designs a detection method of small and weak moving targets in complex scenes based on local clustering of probability and statistics,by establishing space and time domain models,analyzing small target characteristics,background and noise characteristics,extracting small and weak moving target characteristics,and then,using local clustering theory of probability and statistics to divide standard use conditions,realizing target contour curve segmentation,optimizing the whole detection process,and finally completing the design of the detection method.Test results show that,in different experimental images,the average detection rate of the designed weak and small moving target detection method is 69.8%,which has certain practicability and reliability.

关 键 词：概率统计 局部聚类 复杂场景 弱小目标检测 曲线分段 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]