基于3D OGHM的视频运动目标检测算法  

An Approach Based on 3D Orthogonal Gassian-Hermite Mmoments for Detecting Moving Objects

作　　者：李宏友[1] 汪同庆[1] 叶俊勇[1] 蒋昀赟[1] 

机构地区：[1]重庆大学光电技术及系统教育部重点实验室,重庆400044

出　　处：《电子学报》2008年第3期604-608,共5页Acta Electronica Sinica

摘　　要：本文提出了一种新的视频运动目标检测算法.该算法联合了视频序列图像特有的时、空特性,采用3D Gassian-Hermite正交矩法(3D OGHM,3D Orthogonal Gassian-Hermite Moments)进行视频序列中的运动目标检测.并对提出的算法进行了深入的分析和优化,给出了3D OGHM算法的具体实现步骤.实验结果表明:算法能够从复杂场景的视频序列中有效地检测出运动目标;与同类算法相比,不仅有更好的抗噪性能,而且能更有效地增强运动目标信息,且具有较强的鲁棒性和实时性.detecting the Moving objects, we proposed a method based on 3D orthogonal Gassian-Hermite moments （3D OGHM） which combined both the temporal and the special charactersfics of the video sequences. Then we analyzed and optimized our algorithm thoroughly.And the detail steps to implement 3D OGHM on the computer were provided later.Finally, a visual comparison involving similar methods was given. The experimental results show that the moving targets can be effectively detected and extracted from the image sequences of the complex scenes. Compared with the congener methods, ours can not only suppress the noises efficiently but also have a stronger ability to enhance the motion information. And it also has a robust and real-time pefformalice.

关 键 词：视觉跟踪 运动目标检测 Gassian-Hemfite正交矩 信噪比 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]