红外小目标检测的快速并行算法  

Fast parallel algorithm for small infrared target detection

作　　者：汪国有[1] 王岳环[1] 张天序[2] 

机构地区：[1]华中理工大学图像识别与人工智能研究所 [2]图像信息处理与智能控制国家教育部开放实验室

出　　处：《红外与激光工程》1999年第3期13-18,共6页Infrared and Laser Engineering

摘　　要：利用平均灰度绝对差最大测度（ＡＧＡＤＭ）估计目标背景的不相似性，提出了一种红外小目标检测算法，分析了预处理前后图像信噪比增益与目标大小的关系，进而建立了检测概率和信噪比增益的关系，理论上证明了该算法具有优越的性能。接着，在分析了ＭＰＰ并行计算机体系结构的基础上，提出了两种基于ＭＰＰ的小目标检测并行快速实现算法。理论分析表明，基于ＭＰＰ的ＡＧＡＤＭ小目标检测并行快速实现算法运算时间复杂性为Ｏ（Ｌ２ｉ／Ｎ）。In this paper, an algorithm for small infrared target detection using the average gray absolute difference maximum (AGADM) that can compute dissimilarity between an object and a scene is presented, then the signal\|to\|noise gain between original images and their preprocessed results as a function of an object size is analyzed and the theoretical relation between detection probability and the signal\|to\|noise gain is also established, which has shown the performance enhancement of this algorithm theoretically. After analyzing the structure of a MPP computer, two implementations for AGADM\|based small target detection on a parallel MPP computer are presented. Theoretical analysis has shown that the computational complexity for MPP\|based small target detection is O(L\+2\-i/N).\=\;

关 键 词：目标检测 自动目标识别 红外技术 应用 

分 类 号：TN219[电子电信—物理电子学] TN215