摘要
随着人工智能、数字图像处理、模式识别等领域的突飞猛进的发展,智能视频监控日益成为一门应用广泛的综合性学科,其中目标跟踪技术该领域中一个具有重要研究意义的课题。基于传统Mean shift跟踪算法尺度变化易受干扰情况。本文采用一种尺度自适应的Mean shift算法,提出了通过相似度比较更新带宽的新方法。分别计算当前帧和前一帧中中心像素点、边缘像素的相似度函数ρ12,ρ′12,通过比较两者大小判断出目标尺度变化情况,以实现尺度更新。通过试验证明该种方法对于目标尺度更新有更好的效果。
As the development areas of artificial intelligence,digital image processing and pattern recognition,intelligent video monitoring is increasingly becoming a widely used disciplines,in this field,Object tracking with a important significance have a widely research value. Based on the scale change is susceptible in the traditional Mean shift tracking algorithm. In this p, aper,I put forward a scale adaptive Mean shift algorithm. Firstly calculate the similarity function of the central pixels in the current frame and the previous frame and edge pixels, by comparing the size of ρ12,ρ'12 and judging the target scale changes, in order to update the scale. Through the test shows that this method acquire better effect for target scale update.
出处
《国外电子测量技术》
2013年第4期69-72,共4页
Foreign Electronic Measurement Technology