基于傅立叶描述子特征的红外成象GIF目标识别算法及实现  

The Target Recognition Algorithm and Application in Infrared Imaging GIF Based on Fourier Descriptor

作　　者：赖海燕[1] 涂建平[1,2] 周军伟[1] 

机构地区：[1]中国空空导弹研究院,洛阳471009 [2]清华大学精密仪器与机械学系,北京100084

出　　处：《宇航学报》2006年第B12期116-120,131,共6页Journal of Astronautics

摘　　要：在红外成象GIF引信工作过程中，实时目标识别及分类是一个重要的环节，它对于引信在弹道终端参与目标的继续跟踪、瞄准点的选择以及炸点的控制方面有很大的帮助。本文提出一种基于傅立叶描述子特征的红外成象GIF目标识别算法，利用归一化的、表示目标完整轮廓的边界点集的傅立叶描述子特征，结合BP神经网络，完成了巡航导弹、直升机、战斗机三种类型目标的识别。通过MATLAB仿真，验证了该算法的可行性，并在DSP硬件开发平台上对程序代码进行了测试，通过傅立叶描述子特征提取程序的改进，对算法进行优化，使算法的实时性得到较大的改善。As an important part of imagining infrared GIF fuze＇s work, target recognition is very helpful for fuze to take part in continue tracking, estimate the aiming point and control the burst at the endgame of trajectory. An target recognition algorithm for imaging infrared GIF fuze based on Fourier descriptor is brought out in this paper, which made use of the normalized Fourier descriptor about the edge point set that expressing the full contour of target, combined the BP neutral network and implemented the target recognition about missile, copter and fighter. Through the MATLAB program simulation, the availability of algorithm is proved. The test of the program in DSP hardware developing flat is also made. Though the program optimization and improvement, the real - time feature has been improved evidently.

关 键 词：目标识别 红外成象GIF 傅立叶描述子 BP神经网络 

分 类 号：TN911.73[电子电信—通信与信息系统]