摘要
从视频运动背景中检测与跟踪运动目标通常需要对背景进行补偿校正。本文提出了一种利用高性能FPGA实时实现的技术方案。该方法基于仿射参数模型求取视频图像的全局运动参数,利用块匹配的方法得到特征点的运动位移估计,利用匹配加权函数的方法剔除目标点和噪声点对全局运动参数估计的影响。在系统实现上采用多分辨率和并行流水线的处理结构,使系统资源占用和运算效率得到了较好的统一。
The compensation of motion background is always necessary for detection and tracking motion object. A technology that uses high-capability FPGA to compensate motion background is presented. The technology uses affine parameter model to acquire the global motion parameter and block matching method to acquire motion displacement estimation of feature point and matching weighting function to eliminate the influence of objects and noise. In the system based on high-capacity FPGA multi-resolution and parallel pipeline structure is used in order to save system resource and improve system efficiency.
出处
《电子测量与仪器学报》
CSCD
2006年第5期68-72,共5页
Journal of Electronic Measurement and Instrumentation
关键词
全局运动参数
仿射参数模型
块匹配算法
匹配加权函数
多分辨率和并行流水线结构
globe motion parameter, affine parameter model, block matching algorithm, matching weighting function, multi-resolution and parallel pipeline structure.
