摘要
为了动态识别和实时跟踪目标并提高跟踪精度,采用了基于红外亚图像的二阶锁相环跟踪算法。对于图像的识别,传统方法是通过基于二值化块后采用的方形或圆形匹配模板进行目标匹配的,这种方法的匹配速度慢且容易导致匹配失败。本文采用块二值化处理和凝聚性较好的类圆匹配模板,实现了快速跟踪与识别,改善了跟踪的实时效果。实验表明,该方法在图像识别速度上比传统的识别方法要快,且对凝聚性较好的目标的识别性有显著改善,动态跟踪的实时性响应也提高了20%以上。
To identify a target dynamically and track it within an improved accuracy in real time, a second-order phase-locked loop tracking algorithm based on infrared sub-images is used. For image recognition, traditional methods implement target matching by using a square or circular matching template based on block binarization. These methods have a lower matching speed and are liable to result in the failure of matching. In this paper, fast target tracking and identification are implemented by using the block binarization processing and a round matching template. Its real-time performance is improved. The experimental results show that this method is faster than the traditional method in image recognition; its identification performance for the target with better cohesion is improved obviously and its real-time dynamic tracking responsibility is increased by 20%.
出处
《红外》
CAS
2010年第2期25-28,共4页
Infrared
关键词
玫瑰线扫描
亚成像
子块
二值化图像
识别与跟踪
rosette scanning
sub-imaging
sub-block
binarization image
indentification and tracking
