运动点状目标检测的改进算法  

Modified Algorithms for the Detection of Moving Point Targets

作　　者：王绍霖[1] 郑小松[1] 王亦斌 

机构地区：[1]同济大学电气工程系

出　　处：《同济大学学报（自然科学版）》1994年第2期220-224,共5页Journal of Tongji University:Natural Science

摘　　要：讨论了用Ｈｏｕｇｈ变换、匹配搜索和神经网络来检测点状运动目标的方法．当目标信号微弱时，目标检测必须依靠多帧信号的积累．Ｈｏｕｇｈ变换、匹配搜索和神经网络对于可检测信杂比、计算复杂性及可实现性等各有其特点．本文通过计算机仿真和理论分析对几种算法作了比较，并提出了相应改进的综合算法．This paper discusses modified algorithms for moving point target detection, which include Hough Transform, match-searching and neural network algorithms. When the target signal is dim with additive noise, the detection must rely on the accumulation of the signals within several frames of scene images. The algorithms presented here have been modified for less computational complexity and easier real-time realizability under low SNR situations. The comparison of these algorithms is made and the new synthetic algorithms have been demonstrated with computer simulations.

关 键 词：运动点 目标检测 神经网络 匹配 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]