摘要
灰度投影法在灰度值单一、对比度差的图像序列中的运动估计不精准,易造成误差。针对该问题,提出一种基于梯度投影法的电子稳像算法。该算法根据梯度帧内变化剧烈、帧间变化稳定、梯度投影曲线变化明显等特点,利用图像的梯度分布情况进行运动估计。实验结果表明,该算法可提高微光图像序列的帧间运动估计精度。
In view of the problem that the gray projection algorithm can not deal with low contrast ratio image sequences accurately, this paper presents an Electronic Image Stabilization（EIS） algorithm based on gradient projection method. It can utilize the distribution of the image gradient to estimate motion according to the advantage of gradient feature changes violently intra-frame and changes steadily inter-frame. Experimental results demonstrate that the algorithm can effectively improve the motion estimated accuracy under low-light level image sequence.
出处
《计算机工程》
CAS
CSCD
2012年第6期230-232,235,共4页
Computer Engineering
关键词
电子稳像
灰度投影法
梯度投影法
微光图像
光流
Electronic Image Stabilization（EIS）
gray projection method
gradient projection method
low-light level image
optical flow
