【作者】 陈宇；

【导师】 王文生；

【作者基本信息】 长春理工大学 ， 光学工程， 2009， 博士

【摘要】 应用实时联合变换相关器进行复杂背景目标的相关探测与识别,一直目标探测领域的前沿技术。相对于传统的匹配滤波器,联合变换相关器具有处理速度实时、定位精度高等优点,同时又兼有计算机处理的灵活性,可实现实时对于复杂背景下目标的相关探测与识别。由于这一技术有很高的识别率,已被应用于导弹、火箭的导航系统上。本文应用实时联合变换相关器对复杂背景目标进行了相关探测与识别的理论与实验研究,研究结果表明,由于观察视角或观察距离的改变,由图像采集系统采集到的目标图像相对于模板图像存在一定的角度畸变和比例畸变,而传统的联合变换相关器对于各种形式的畸变都是非常敏感的,很大程度上影响了目标的识别率。本文研究了多种解决畸变不变目标识别的算法及其优缺点,其中对极坐标-梅林变换法进行了深入研究,给出了光学实验结果。经过分析与比较,最终确定选用最大平均相关高度(MACH)滤波器进行功率谱面滤波。编制了MACH滤波器图像处理程序,并由其谱面滤波改进为对功率谱滤波,使其成功应用于联合变相关器,进行了光学相关目标识别的模拟实验和光学实验。实验结果表明,应用改进后MACH滤波器程序,可以有效增强相关点亮度,实现复杂背景静态目标畸变不变的目标相关探测与识别。针对复杂背景动态目标的识别与跟踪问题,提出了动态模板技术,把运动目标的瞬时状态作为模板,并结合输入面阈值处理方法进行了光学实验研究,实验结果表明,应用该方法可有效解决复杂背景动态目标的识别与跟踪问题。更多还原

【Abstract】 Detection and recognition of targets with cluttered background by real-time joint transform correlator is always the front technology in this field. Compared with conventional matched filter, joint transform correlator has many advantages such as real-time processing speed, high positioning accuracy and the flexibility of computer process, which can satisfy the requirement of detection and recognition of targets with cluttered background. This technique has been applied in navigation systems of missiles and rockets for its high recognition rate.Theoretical and experimental research is developed on detection and recognition of targets with cluttered background by real-time joint transform correlator in this paper. Research result shows when observation perspective or observation distance changes, angular or scale distion appears, which is produced by target image relative to the template image. While conventional joint transform correlator is very sensitive to various distortions, which effect the recognition rate to a great extent. Several algorithms and their advantages and disadvantages are researched in this paper to solve the recognition problem of distortion invariant targets, in which polar-in transform method is researched deeply and the experimental results are given. After analysis and comparision, Maximum average correlation height(MACH) filter is adopted to realize filtering in power spectrum plane.Computer programs of MACH for image processing are compiled. To make it applied in joint transform correlator, the spectrum filtering is changed to power spectrum filtering in the program. Simulation experiments and optical experiments of optical correlation recognition are conducted. The results show the intensity of correlation peaks are strengthened and distortion invariant target detection and recognition in cluttered background is realized after MACH filtering.As to the recognition and tracking of moving targets with cluttered background, dynamic template technology is proposed, which is to take the instantaneous state of moving targets as the template. Optical experimental research is conducted by dynamic template technology combined with threshold processing in the input image plane. Research shows the recognition and tracking of moving targets with cluttered background can be realized.更多还原