视频序列中二值图像的快速目标检测方法  

Fast Object Detection Method for Binary Image in Video Sequence

作　　者：李国辉[1] 焦波[1] 涂丹[1] 李燃[2] 汪彦明[1] 

机构地区：[1]国防科技大学信息系统与管理学院,长沙410073 [2]总参工程兵科研三所,洛阳471023

出　　处：《计算机工程》2009年第18期217-218,226,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(60273066)

摘　　要：针对视频序列中二值图像目标检测时间效率不稳定的问题,提出一种快速目标检测方法,其主要步骤包括逐层膨胀、填补空洞、逐层腐蚀和检测目标。理论分析与实验结果证明,该算法的时间效率、时间稳定性和检测质量较高,其最差时间效率与采用3×3结构元素的形态学闭算子相当。Aiming at the instability problem of the object detection time efficiency of binary image in video sequence, this paper proposes a fast object detection method. The main steps of this method include dilating layer by layer, filling holes, eroding layer by layer and detecting objects. Theory analysis and experimental results confirm high time efficiency, time stability and detection quality of this method, and its worst time efficiency is approximately equal to 3 ×3 structuring element＇s morphological closing operator＇s time efficiency.

关 键 词：快速目标检测 二值图像 视频序列 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]