提出了一种适合Bayer彩色序列图像的电子稳像方法。该方法根据Bayer彩色图像特点, 通过色彩还原和RGB至YCbCr颜色空间的变换, 将Bayer彩色图像转换为灰度图像。采用灰度投影算法估计两幅图像之间的运动量; 通过计算灰度图像在水平和垂直方向的投影特征数据, 构造当前图像和参考图像投影数据之间的相关函数; 依据最小误差匹配准则, 估计两幅图像之间的水平和垂直位移量。为解决传统补偿算法在处理奇数运动量时的颜色失真问题, 基于双线性插值算法构建了适合Bayer彩色图像的运动补偿算法。最后, 对Bayer彩色序列样本图像进行了色彩还原和图像稳定实验。结果表明, 本文方法可以实现像元级运动量检测, 运动补偿后的序列图像峰值信噪比优于40 db。实验显示, 本文方法可行且有效,适合Bayer彩色序列图像的运动估计和运动补偿, 可以实现序列图像的稳定输出。

A novel method for Electronic Image Stabilization (EIS) of Bayer color video image sequence was proposed. On the basis of the characteristics of Bayer color image, it converted firstly the Bayer color image into a gray image by using demosaicing and converting RGB into YCbCr. Then, the motion estimation algorithm based on gray scale projection was used to estimate the translation parameter between current frame and reference frame, in which the gray scale projection curves of horizontal and vertical directions of the gray image well reflected the feature of gray distribution in the corresponding direction, and the translation parameters could be obtained via cross correlation function of the gray scale projection curves. In order to resolve the problems of mismatch-color when translation was odd, the motion compensation algorithm based on the bilinear interpolation was applied to compensation of the translation of current frame. Finally, the demosaicing and image stabilization for the Bayer image sequence were performed. The experimental results for video sequences show that the proposed method can implement image detection in a pixel level, and can achieve higher stabilization accuracy with a peek signal to noise ratio of 40 db. It is efficient and suited for the motion estimation and compensation for Bayer color video image sequences.