地面抛撒地雷红外成像检测与识别技术

发布时间：2019-03-30 08:02

【摘要】：红外成像技术在军事目标检测与识别中应用越来越广泛。基于前视红外成像的探雷技术具有探测成像分辨率高、视场范围大、效率高、精度高等优点,已成为世界各国成像探雷技术研究的热点。本文以地面抛撒地雷探测为目标,按照系统设计与实现的流程,对红外成像技术在探雷系统中的应用进行分析,主要从如下几个方面展开研究工作:(1)从红外成像技术应用于探雷系统的基本原理出发,进行可行性分析;然后结合地面抛撒地雷特征,通过红外辐射数学建模实现不同环境条件的影响因素分析,并给出几种特殊的影响因素。(2)搭建基于车载的红外成像探雷系统,从系统组成、基本工作原理以及系统工作流程进行分析,并设计算法处理流程实现地雷目标的检测、识别及定位。(3)根据红外成像焦平面阵列可能存在的非均匀性噪声,选取列均衡化的非均匀性校正算法,复原降质的红外图像;再运用自适应分段线性变换算法,达到增强地雷目标抑制复杂背景的目的,且在处理效率上满足系统实时性要求。(4)结合预处理后的红外地雷图像,从基于阈值的分割方法出发,根据红外布雷场景灰度分布特征,选取交叉熵约束的阈值分割方法,与传统的类间方差或最大熵方法相比,分割的地雷目标更准确;为了后期地雷目标的精确识别,还对阈值分割后的图像进行背景噪声消除和地雷目标空洞填充。(5)利用红外布雷场景分割后的二值图像,统计不同待识别地雷目标的连通域;根据特征不变量的红外图像目标识别原理,建立似圆形地雷目标的凸壳不变量模型,求解待识别目标区域的凸壳不变量,并进行匹配与筛选;最后利用灰度重心定位算法实现地雷目标位置坐标的标识。实验结果表明,本文选取的算法处理流程不仅能够实现红外布雷场景中地雷目标的高精度探测,而且满足红外成像探雷系统实时性处理要求。
[Abstract]:Infrared imaging technology is more and more widely used in military target detection and recognition. Mine detection technology based on forward-looking infrared imaging has the advantages of high resolution, wide field of view, high efficiency and high precision. It has become a hot spot in the research of imaging mine detection technology all over the world. According to the flow of system design and realization, the application of infrared imaging technology in mine detection system is analyzed in this paper. Mainly from the following aspects of research work: (1) from the infrared imaging technology applied to the basic principles of mine detection system, the feasibility analysis; Then combined with the characteristics of ground-thrown mines, the influence factors of different environmental conditions are analyzed by mathematical modeling of infrared radiation, and several special factors are given. (2) the infrared imaging mine detection system based on vehicle is built, which is composed of the system. The basic working principle and the workflow of the system are analyzed, and the algorithm processing flow is designed to realize the detection, recognition and location of mine targets. (3) according to the possible non-uniform noise of infrared imaging focal plane array, The non-uniformity correction algorithm of column equalization is selected to recover the degraded infrared image. Then the adaptive piecewise linear transformation algorithm is used to enhance the mine target suppression complex background, and the processing efficiency meets the real-time requirements of the system. (4) combined with the pre-processed infrared mine image, Starting from the threshold-based segmentation method, according to the gray distribution characteristics of infrared mine-laying scene, the threshold segmentation method with cross-entropy constraint is selected. Compared with the traditional inter-class variance or maximum entropy method, the mine target segmentation is more accurate. In order to accurately identify the mine target in the later stage, the background noise and the hole filling of the mine target are removed from the threshold segmented image. (5) the binary image of the infrared mine-laying scene is used to segment the image. (B) Statistics the connectivity areas of different mine targets to be identified; According to the principle of infrared image target recognition with feature invariants, the convex hull invariants model of similar circular mine targets is established, and the convex hull invariants in the target region to be identified are solved, and the matching and screening are carried out. Finally, the gray center of gravity localization algorithm is used to mark the location coordinates of mine targets. The experimental results show that the proposed algorithm can not only realize the high precision detection of mine targets in the infrared mine-laying scene, but also meet the real-time processing requirements of infrared imaging mine detection system.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.41;TN219

【参考文献】