前视红外末制导Nprod目标识别算法适应性研究  

Research on Applicability of Nprod Target Recognition Algorithm for FLIR Terminal Guidance

作　　者：刘思雨 王雪梅 杨小冈[1] 马治明[2] 朱瑞奇 

机构地区：[1]第二炮兵工程大学三系 [2]中国人民解放军96111部队

出　　处：《科学技术与工程》2013年第15期4375-4378,4383,共5页Science Technology and Engineering

基　　金：国家自然科学基金(61203189);总装预研基金(9140A01060411JB4701)资助

摘　　要：描述了前视红外末制导景象匹配中归一化积相关(Nprod)目标识别算法的基本原理。由实际拍摄的典型目标前视红外图像匹配结果得出:当目标所占面积越大、与背景灰度反差越大、边缘越清晰,Nprod算法匹配效果越好;在目标与背景的灰度分布集中或者目标较远的情况下,匹配效果不甚理想。给出了Nprod目标识别算法的基准图预处理方法,对远距离情况下建筑物目标的匹配进行了匹配方法研究,实验表明具有较好的匹配效果。Describe the basic principle of Normalized product correlation （Nprod） target recognition algorithm in the forward-looking infrared （FLIR） terminal guidance scene matching. FLIR image matching results in the actual shooting of the typical targets show： the larger the target occupies the area, the greater the background grayscale contrast, and the sharper the edge, then the better the Nprod algorithm matches the results. When the distribution of gray-scale of the target and background is concentrated, or the target is far away, the result is not very satisfactory. The reference images preprocessing method for Nprod target recognition algorithm has been given and the matching method for long-distance buildings target, the experiment shows a good result.

关 键 词：归一化积相关算法 前视红外末制导 景象匹配 基准图 实时图 

分 类 号：TJ765.3[兵器科学与技术—武器系统与运用工程]