摘要
提出了隔帧差分向量无穷范数检测运动目标的数学模型,建立了能够充分发挥该模型能力的并行处理结构体系。由于算法以隔帧差分为基础,所以处理系统不仅能够探测到帧间位移不小于1个像元的点目标,而且可以探测到帧间位移小于1个像元而多帧累积位移大于1个像元的运动点目标,使算法探测与识别目标的能力大大提高。实验结果表明了算法的有效性。
A mathematical model to detect the moving targets by the discontinuous frame difference vector's infinite norm is proposed. Then the parallel processing construction system is established, which can make full use of the moving point targets detection ability of the discontinuous frame difference vector's infinite norm. Due to the algorithm is based on the discontinuous frame difference, the parallel processing system can detect not only the moving point targets whose displacement is no less than one pixel between two continuous frames, but also the moving point targets whose displacement is no more than one pixel between two continuous frames but larger than one pixel between two discontinuous frames. Therefore, the ability of detecting and recognizing the moving point targets is highly improved. The experimental result shows that the algorithm is efficient. \;
出处
《红外与激光工程》
EI
CSCD
北大核心
2003年第2期157-162,共6页
Infrared and Laser Engineering
基金
中国科学院创新基金项目与"十五"资助项目(863 2002AA731264)
关键词
图像分割
点目标
检测
并行处理
Image segmentation
Point target
Detection
Parallel processing
