摘要
针对灰度投影算法在稳像时用时较长的问题,提出一种快速三点搜索算法。在不影响稳像效果的前提下,对图像进行分区,用改进的算法进行灰度投影的矢量求解,并运用均值滤波得到全局运动矢量,实现运动补偿。实验结果表明,该算法计算的峰值信噪比(PSNR)与传统的灰度投影算法比相差小于0.5dB,但时间却减少了21.8ms。
Aiming at the gray proj ection algorithm for image stabilization with a relatively long time question,a new algo-rithm based on three-point search is proposed.The image is divided into several blocks without affecting the image stabiliza-tion effect,the algorithm is used for solving gray proj ection vector.Finally,the global motion vector is obtained by using the mean filter,the motion compensation is achieved.The experimental results show that the PSNR is less than 0.5 dB by the proposed algorithm than the traditional gray proj ection algorithm,however the time is reduce by 2 1 .8 ms.
出处
《桂林电子科技大学学报》
2015年第2期152-155,共4页
Journal of Guilin University of Electronic Technology
基金
广西自然科学基金(2013GXNSFCAA019330)
广西可信软件重点实验室基金(kx201213)
广西教育厅科研项目(2013YB086)
关键词
灰度投影算法
运动估计
三点搜索
gray proj ection algorithm
motion estimation
three-point search
