一种基于特征跟踪的图像稳定方法
A Feature-Tracking Based Method for Image Stabilization
-
摘要: 根据摄像机透视投影模型,通过提取特征点进行摄像机的运动估计,提出一种基于FMA(frame-to-mosaicalgorithm)的补偿方法.该方法针对图像连续帧间晃动幅度较大的情况,通过均值滤波合成当前帧及其前几帧的运动进行补偿.实验结果表明该方法有较好的稳定效果,而且能够做到实时处理.Abstract: A motion model is presented to estimate the motion parameters for image stabilization by tracking feature points from consecutive frames. A frame to mosaic algorithm (FMA) for motion compensation is described. The proposed algorithm is able to compensate the motion by fusing the motion of the current frame with that of the previous frames. Experimental results show its effective and potential applications.
下载: