群发炮弹炸点自动识别与定位技术研究

发布时间：2018-02-28 05:35

本文关键词： 群发炮弹炸点识别与定位红外双目视觉目标检测与跟踪空间位置坐标解算　出处：《华中科技大学》2016年硕士论文　论文类型：学位论文

【摘要】：群发炮弹炸点实时自动识别与定位技术在炮兵军事训练评估系统中有广泛应用,传统的人工定位方法已不能满足现代化数字靶场的军事需求。论文针对群发炮弹炸点实时精确识别与定位的实际应用需求,提出了一种基于红外成像的双目视觉群发炮弹炸点自动识别与定位系统解决方案(PZC系统),并对单目视场炮弹炸点自动检测、双目视场同名炸点自动匹配等关键技术开展了深入系统研究,相关成果已集成在PZC系统中,并通过多次靶场试验验证了方案的可行性。论文主要工作如下:论文结合国内外相关文献资料,在已有研究的基础上,针对群发炮弹炸点目标特征和红外图像特征,提出了基于红外双目视觉的群发炸点自动检测与定位的一个系统集成解决方案。该解决方案通过在靶场架设左右红外观测站获取红外图像序列,并对采集到的红外图像进行预处理,在处理过后的红外图像序列中进行群发炸点目标实时自动识别和定位。论文提出了红外观测站自动曝光补偿控制解决方案,结合边缘检测使用反锐化掩模算法并应用于红外图像细节增强上,效果显著。炸点目标因其固有特征,沿用传统的目标检测算法检测效果不理想,论文通过对VIBE(可视化背景建模,Visual Background Extractor)算法进行了研究,提出了基于VIBE和炸点目标特性的炸点目标检测方法,设计了消除目标内部碎片后处理方法。针对炸点目标的个体特征和群体特征,论文提出了基于目标链和目标时空序列特征的红外群发炸点关联跟踪算法,实现序列图像中群发炸点目标的关联与跟踪。论文分析并研究了群发炸点目标双目匹配的特征和难点,提出了基于时间片分割以及双向验证最短公垂线约束的群发炸点双目立体匹配算法,并对匹配的炸点采用基于公垂线中点法解算其三维空间物理坐标。论文基于OpenCV2.4.9在VS2010平台编程实现PZC系统关键算法,并通过对靶场实弹演习采集的视频图像和外场模拟试验采集的视频图像进行了实验分析,综合验证了本文解决方案的实用性和有效性。
[Abstract]:The real-time automatic identification and location technology of the firing point of artillery shells has been widely used in the artillery military training and evaluation system. The traditional manual positioning method can not meet the military requirement of modern digital shooting range. This paper presents a solution to the automatic identification and location system of the burst point of binocular vision group projectile based on infrared imaging. The PZC system is used to detect the explosion point of the projectile in a single visual field. Some key technologies, such as automatic matching of burst points with the same name in binocular field of view, have been deeply and systematically studied, and the related results have been integrated into PZC system. The main work of this paper is as follows: combining with the domestic and foreign literature, based on the existing research, aiming at the target features and infrared image features of mass artillery shells, This paper presents a system integration solution for automatic detection and location of burst points based on infrared binocular vision, which obtains infrared image sequences by setting up the right and left infrared observation stations in the shooting range. The infrared image is preprocessed, and the group burst point target is automatically identified and located in the processed infrared image sequence. A solution of automatic exposure compensation control for infrared observation station is proposed in this paper. Combining edge detection with de-sharpening mask algorithm and applied to infrared image detail enhancement, the effect is remarkable. Because of its inherent characteristics, the detection effect of the traditional target detection algorithm is not ideal. In this paper, the visual Background tractor (VBE) algorithm is studied, and a method of burst point target detection based on VIBE and exploding point target characteristics is proposed. A post-processing method to eliminate the internal debris of the target is designed. Aiming at the individual and population characteristics of the exploding point, an infrared burst point association tracking algorithm based on the characteristics of the target chain and the space-time sequence of the target is proposed in this paper. In this paper, we analyze and study the characteristics and difficulties of binocular matching of group burst point targets. A binocular stereo matching algorithm for group burst points based on time slice segmentation and bidirectional verification of the shortest common vertical constraint is proposed. The method based on the common vertical line is used to calculate the three-dimensional physical coordinates. The key algorithm of PZC system is realized by programming on VS2010 platform based on OpenCV2.4.9. The practicability and validity of the solution are verified by the experimental analysis of the video images collected by the range live ammunition exercises and the video images collected by the external field simulation test.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TJ306;TP391.41

【相似文献】