长波红外主被动复合成像中目标检测仿真研究

Target detection emulation on long-wave infrared active-passive imaging

给出了Jeffrey H.Shapiro等人的目标检测算法,以及在该算法基础上推导出的七个目标检测公式。自行编程设计和实现了仿真实验。在理想情况下,通过仿真对检测性能进行了对比,从仿真结果可知,!的取值在5 dB附近或者大于5 dB时,在激光雷达的工作范围内(CNR≈10￣30 dB),综合利用主被动两种信息进行目标检测的性能要优于仅利用其中任何一种信息进行目标检测的性能,综合利用主被动三种信息进行目标检测的性能要优于仅利用其中任何一种信息以及两两信息组合进行目标检测的性能。强度像、距离像与被动红外融合时的检测性能最高,从而验证了长波红外主被动复合成像的优越性。同时,讨论了理想情况下子帧大小对目标检测公式性能的影响,对非理想情况进行了初步探讨。

The target detection algorithm and seven target detection expressions presented by Jeffrey H. Shapiro were introduced. Then, in the paper the long-wave infrared active-passive emulations were finished. If the value of ξ is near 5 dB or over 5 dB, the detectability using active and passive images to detect target was better than only using single one; the performance of the target detection by using three kinds images was better than only using single one or compounding two images. The performance of jointing intensity image, range image and passive infrared image was the best, so the superiority of longwavelength infrared active and passive images could be proved. On ideal condition, the effect of the size of sub-frame on the performance of the target detection expression was analyzed. Nonideal condition was also discussed preliminarily.