基于ASM实现视频中物体的定位

Localization of Object in Color Video Sequences Based on ASM

对目标进行有效跟踪是目前许多国家获取信息的重要手段,主动形状模型(ASM)主要应用于灰度图像中物体的跟踪与定位,本文针对Koschan等提出的结合金字塔模型的ASM方法作进一步改进,使之应用于视频(本文视频采用MPEG数字视频压缩编码标准)中物体的定位:首先将视频解码还原成图像序列,从人眼视觉特征出发,将RGB颜色模型转换为HSV颜色模型,用色调信息取代灰度信息进行物体定位;并研究图像序列帧之间物体的重合度对提高定位效率的作用。实验结果说明在ASM中使用色调信息能有效地实现视频中的物体定位。

At present tracking objects effectively is an important method of many countries to gain information. Traditionally Active Shape Model is mostly applied to track an object in a gray- level image. In this paper, we present an extension of ASM for color video tracking based on the research of Koschan which implements hierarchical technique in image pyramid. First of all, color video sequences are obtained based on the digital video frequency compression code standard. Then, according to human vision character RGB color space is converted to HSV color space so that the hue information is used for localization of an object instead of the gray information. At last, the tracking performance of superposition of an object in successive frames is studied. Experiments demonstrate that the ASM method combining with the hue information produces effective results for localization in video sequences.