摘要：随着红外成像技术的发展，红外成像技术检测在军事及国民经济中的应用十分广泛。这时，对检测的精度，实时性及智能化提出了更高的要求。面对这一问题，本文基于局部边缘相似性结构模型研究了一种智能的红外成像目标检测方法。

  针对红外图像进行目标检测，本文采用一种自适应回归核(Locally Adaptive Regression Kernels，LARK)的特征提取方法，对红外图像的局部结构特征进行提取.通过去除相似结构进行降维，获得较简洁的局部特征值矩阵。再通过局部匹配，统计测试图片局部区域内包含与模板集相似的局部结构的数量，从而生成测试图像与目标的相似性图片，然后通过非极大值抑制对相似度信息进行进一步提取，最后从目标红外图像中获得与模板相似的目标。仿真结果表明，运用此种方法能够在复杂环境条件下检测出目标，且精度较高。

关键字：红外成像目标  LARK特征提取  显著分析   相似性匹配

毕业设计说明书外文摘要

Title        Infrared imaging target detection              

Abstract：With the development of infrared imaging technology, infrared imaging technology in the military and national economy in the application of a wide range. At this time, the detection accuracy, real-time and intelligent put forward higher requirements. In the face of this problem, this paper studies a method based on the local edge similarity structure model of infrared imaging target detection method

.In this paper, a feature extraction method of Locally Adaptive Regression Kernels (LARK) is used to extract the local structural features of infrared images. Using the infrared image to build the template set, by removing the similar structure to reduce the dimension, get a simple matrix. And then through the local matching, through the statistical test picture local area contains a template set similar to the number of local structure, resulting in test image and target similarity picture, and then through the non-maximum suppression of similarity information to further extract, and finally Obtain a similar goal from the target infrared image to the template. The simulation results show that this method can detect the target under complex environmental conditions, and the precision is high.

Keywords:  Infrared imaging target   Feature extraction Significant analysis   Similarity matching

目   次

1绪论..1

1.1研究背景及检测思路.1

1.2红外成像目标检测的研究现状..3

1.3本论文的主要内容4

2 局部自适应回归核提取特征值5

2.1核回归函数基本理论.5

2.2 采用核回归方法对特征值进行提取7

2.2.1自适应回归核提取特征值的原理.8

2.2.2自适应回归核提取特征（lark）权值矩阵..9

2.2.3基于LARK特征提取方法对红外目标进行检测.10

2.3 本章小结10

3 特征向量降维以及相似性匹配..11

3.1 简单模板集的建立.11

3.2特征向量的降维.12

3.3相似性匹配..13

3.3.1 匹配方法.13  

3.3.2相似度图像的生成15  

3.3.3非极大值抑制获取图像信息.17

3.4本章小结.20

4检测参数与实验结果..20

4.1参数分析.21

4.1.1相似度阈值22  

4.1.2相似度全局阈值T 23  

4.1.3非极大值抑制参数h.23

4.2实验结果展示23  

4.2.1单模板实验效果图23

4.2.2 多模板实验效果图..24

4.3本章小结.25

5实验的不足与展望.26

论文总结

致谢

参考文献