基于梯度和颜色直方图相融合方法的人脸跟踪

利用长短轴比例固定为1.4的椭圆模型模拟人脸,并融合梯度模型和颜色直方图模型进行人脸跟踪,给出了梯度模型和颜色直方图匹配度计算方法.因为两种模型之间在考虑的点集和利用的信息上具有互补性,所以提高了系统的鲁棒性.带加速度的运动预测则有效地减小了检测区域,提高了系统的速度.为降低光照变化的影响,使用一种简单的自适应亮度补偿算法,在光照较弱的条件下对待测图像进行补偿,从而提高了系统的适应性和实用性.实验表明:该方法在复杂背景中甚至在较暗的光照条件下都能取得较好的效果, 可以跟踪人脸任意方向的运动,速度可以达到20帧/s以上.

Gradient and color histogram fusing based head-tracking

This method of head-tracking uses ellipses to simulate human heads, the proportion of whose major axis and minor axis is fixed to 1.4, and it fuses intensity gradient model and color histogram model to track human heads, and the method of calculating the similarity between the template and the image to be detected using the two models is also given. Because the two models are complement to each other as to the element set concerned and the information used, this method improves the robustness of the system. Motion prediction taking acceleration into account could effectively reduce the detection area and it consequently increases the tracking speed. To eliminate the influence of lightness changing, it puts forward an algorithm of adaptive lightness compensation to compensate the image to be detected in faint lighting conditions. Experiments show that it could work well in complex background and even in bad lighting conditions, and it could track arbitrary movement of human heads, and the tracking speed could be above 20 frames per second.