摘要
低信噪比条件下图像序列中运动点目标的检测问题是多维图像处理与目标识别领域的一个难点。本文首先给出了该项研究的前提条件，指出了问题的主要矛盾；然后针对序贯假设检测的基本方法，提出了一个改进的实现框架；最后针对仍存在的问题，提出了对序列先进行分段，并对分段复合，再利用复合帧进行序贯检测的方法。这里引入了分段复合技术，给出一种分段复合方法。该算法层次分明，时空复杂度适中，是国内性能较优的一种算法。
The automatic detection of moving point target in the low SNR image sequences belongs to the difficult in the area of multidimensional signa1(image) processing and pattern recognitionly.Firstly,the premises of prcsent research are reviewed, and some keys to problem are analyzed.Then by introducing the idea of sequential hypothesis testing, the paper puts forward a improved implementing framework. Finally, a new total solution is given-integrating stage before sequential detection.Here the technique of stage integration is definitely submitted with a method for implementing.This algorithm has clear hierarchy, moderate complexity, and better performance.
出处
《系统工程与电子技术》
EI
CSCD
1997年第10期1-5,42,共6页
Systems Engineering and Electronics
基金
国家863计划
关键词
自动目标识别
序贯假设检测
图像处理
Automatic target recognition(ATR), Point target, Low SNR, Sequential hypothesis testing, Stage integration.
