摘要
针对星载光电成像系统的空间目标检测与识别技术,设计了一套数字信号处理器（DSP）结合现场可编程门阵列（FPGA）的图像处理系统,并提出了一种卫星目标检测与识别方法。在没有先验知识条件下,利用改进的图像差分法对具有星空背景的序列图像进行目标检测,再根据卫星与天然星体具有不同的边缘轮廓信息,通过最小距离分类器识别出卫星目标,获得卫星模板,从而可进行模板匹配以实现卫星的快速检测和识别。实验结果表明,提出的方法相对于传统方法,运算时间减少了15%-40%,满足星载光电成像系统对卫星目标检测与识别的准确性与实时性的要求,在工程实践中具有一定的应用价值。
In order to achieve the space target detection and identification based on the satellite photoelectric imaging system, an image processing system is designed with combination of a leading processor digital signal processor （DSP） and an auxiliary processor field programmable gate array （FPGA）. A method of the satellite targets recognition is put forward. The targets are detected with the modified differential method in sequential images on the starry background, in the absence of a priori knowledge conditions. The satellites are identified with the minimum distance classifier and their templates are acquired based on the different profiles of the satellites and natural stars. Therefore, the rapid detection and identification of the satellite targets can be achieved via the template matching algorithm. Experimental results show that compared with traditional method, the running time is reduced by 15%-40%. It satisfies the accuracy and real-time requirement of satellite target detection and identification with practical value.
出处
《激光与光电子学进展》
CSCD
北大核心
2014年第12期112-118,共7页
Laser & Optoelectronics Progress
关键词
图像处理
目标检测
目标识别
image processing
target detection
target identification
