多颜色标记三维定位以及标记点稳定性研究

Research of 3D Position Detection and Stability Based on Multicolor Marks

在基于颜色标记的手部识别以及运动跟踪的过程中,由于背景及噪声的干扰,二维图像点坐标会发生波动,产生误差,导致由二维图像坐标计算手部三维坐标的过程中,静止状态下的带标记的目标在虚拟环境中会发生抖动。采用自适应分块区域增长法跟踪颜色标记,获得标记点区域范围,在该范围内进一步采用局部Hough变换精确定位颜色标记。实验证明,该方法在不降低亚像素精度的前提下,可以稳定定位手部标记点,且具有先验知识的局部Hough变换,在克服已有Hough变换不足的同时,能有效提高定位精度。

During hand detection and motion tracking based on color marks,the background color and noise will disturb image inevitably in common environment and the position of color marks in 2D image will be unstable.Thus the 3D position error calculated from 2D images will be magnified even the target marks are fixed.A method was proposed to ameliorate this situation.The region grow pixel was used to find and track the region with mark,then the Hough transform was applied to aim the color precisely in the mark region.Experimental results showed that the proposed algorithm can stable the marks position without reducing the sub-pixel precision.Using Hough transform with known information,will not burden the entire recognizing and tracking efficiency unless losing target in the image.