摘要
针对视频图像中目标形态的检测识别问题,提出了一种基于凸型模板修正算法对条形目标的检测、识别和定位技术。通过分析凸型模板识别条形目标的基本原理,提出引入倾斜角度因子的凸型模板改进算法,提高了条形目标识别的准确度,并以单人个体作为研究对象进行实验,结果表明这种方法运算复杂度低,对条形目标的识别率高,跟踪特性良好。
The paper introduces a fast object tracking and behavior recognition technology applicable for human detection and gesture recognition in video and image processing filed. It analyzes the principle of torso mask and implementation steps of the improved method. This method increases the accuracy of torso location through introducing torso tilt angle factor. The experimental results show that the new method achieves lower complexity in calculation, higher recognition in object and better performance in tracking.
出处
《计算机与网络》
2011年第14期49-52,共4页
Computer & Network
