论文部分内容阅读
美国圣地亚国家实验室的研究人员研制出一种供导弹制导用的光电子基图像识别器。这种声光图像相关器使用光来处理并传输信息。光电子基系统由于光学系统固有的并行性和光波的高信息容量,具有极高的计算速度,而且功耗低于电子基系统。图像相关器只需要约100W的功率并且尺寸大约象桌面那样大。为了验证系统,研究人员使用一种原理型混合光学处理机。卡车和装甲车模型的视频图像利用声光器件用电子方法扫描到相关器中。系统有256个代表待识别物体的数字式模式,叫做模板。它们存贮在数字存储器中,并在数模
Researchers at Sandica National Laboratory in the United States have developed a photoelectron based image recognizer for missile guidance. This acousto-optic image correlator uses light to process and transmit information. Optoelectronic-based systems have an extremely high computational speed due to the inherent parallelism of optical systems and the high information capacity of light waves, and consume less power than electronic based systems. The image correlator only requires about 100W of power and is about as large as the desktop. To validate the system, researchers used a prototype hybrid optical processor. Video images of truck and armored vehicle models are digitally scanned into correlators using acousto-optical devices. The system has 256 digital representations of objects to be identified, called templates. They are stored in digital memory and are in digital form