This paper proposes an object-oriented electronic image stabilization algorithm based on different depths of field. It uses Harris angular point detection algorithm obtained at different depths of field goal of characteristic points. According to the feature point sets inside structure tight teams constraints and distance constant property on criterion, it validates matching feature points, and works out the motion vector equation to gain motion vectors in different depths of field range. Tracking differentiator filtering is used to remove jitter interference, and the objects in different depths of field range are compensated separately. Experimental results show that the algorithm can solve stabilization problem for objects in different depths of field.%提出一种针对不同景深目标的电子稳像算法.利用Harris角点检测算法获得处于不同景深的目标特征点,根据特征点集内结构紧支性约束及距离不变准则,对特征点匹配进行验证,通过求解运动矢量方程,得到处于不同景深范围的运动矢量,使用跟踪微分器滤波去除抖动干扰,对处于不同景深的目标分别进行补偿.实验结果表明,与同类算法相比,该算法的图像补偿效果更好.
展开▼
