摘要
图像中待攻击目标的自动识别与跟踪锁定是图像制导的基本任务,也是图像制导技术的难点所在。由于获得图像的角度和位置的不同,同一目标在不同图像中会有较大的变形,如何正确地识别出该目标是至今未能很好解决的问题。相比之下,人和动物在复杂的环境中,几乎在各种角度和位置的情况下,均可以可靠地识别出目标,并能完成对该目标的跟踪锁定。因此生物视觉的工作机理成为了当前学者研究的热点,在继承和发展了当前国际上该领域的最新研究成果的基础上,实现了生物视觉原理下的目标识别与跟踪算法,经过大量真实图像的试验,证明了该算法的鲁棒性。
The target recognition and tracking in image is the basic task and the cnoge point in image-guidance technology. Because of different position and angle of view, the same object may have serious deformation in images, so how to recognize the one is not well resolved. Compara- tively, the people and animals, almost at all angle and position of view, it is easy to recognize and track it, so biological vision becomes a hot research domain. Inherit the newest research result in biological vision, the algorithm based on biological vision to recognize and track the target is implemented. After tested by many frames of real image, it can be shown that the algorithm is robust.
出处
《红外与激光工程》
EI
CSCD
北大核心
2007年第6期977-979,995,共4页
Infrared and Laser Engineering
关键词
生物视觉
感受野
仿射模型
目标识别
目标跟踪
Biological vision
Receptive field
Affine model
Target recognition
Target tracking