摘要
结合相关匹配算法特点提出了一种改进的基于可见光图像制备的参考模板进行二值匹配的目标检测识别算法,算法包括实时图像分辨率、同质变换、OTSU聚类、结构匹配定位等一系列处理过程。所提出的算法结构简单、计算量小、易于并行实现,在基于TMS320C6713的多DSP系统上处理一帧实时图像仅耗时14ms,可以应用于工程实际之中。
According to the characteristic of the correlation matching algorithm, a improved matching algorithm based on referenced model prepared by visible light image is proposed. This algorithm consists of a series of transaction including real-time image resolution reducing, homogeneity transform, OTSU clustering, structure match orientation etc. It has the characteristics as following： simple structure,low consume, easy parallel realization etc. Its time consume processing a frame infrared image in multi-DSP system based on TMS320C6713 is only 14ms,so it can be applied to the engineering item.
出处
《激光与红外》
CAS
CSCD
北大核心
2007年第12期1322-1324,共3页
Laser & Infrared
基金
国家自然科学基金项目(60135020FF030405)资助
关键词
图像处理系统
实时性
自动目标识别
相关匹配
参考模板
image processing system
real-time performance
automatic target recognition
correlation matching
referenced template
