一种新的卡尔曼滤波图像复原算法

A New Kalman Filter Image Restoration Algorithm

为了更好地对退化图像进行降噪,提升算法运行效率,同时保留复原图像边缘细节信息,提出了一种交替卡尔曼滤波图像复原算法。与传统方法相比,该算法不需要估计退化函数,仅需将退化图像的信息矩阵第一行(或列)代入卡尔曼滤波预测方程作为初始值进行滤波,然后将第一次滤波图像的信息矩阵第一列(或行)代入卡尔曼滤波预测方程作为初始值进行第二次滤波获得复原图像。仿真结果表明,交替卡尔曼滤波图像复原算法在去除退化噪声保证图像清晰的同时,还能快速完成图像复原,其计算时间降为维纳滤波复原法计算时间的2%。

In order to reduce the noise of the degraded image better, improve the efficiency of the algorithm, and retain the details of the edge of the restored image, an alternative Kalman filter image restoration algorithm is proposed. Compared with the traditional method, this algorithm does not need to estimate the degradation function. It only needs to substitute the first row(or column) of the information matrix of the degraded image into the Kalman filter prediction equation as the initial value for filtering, and then filter the image information for the first time. The first column(or row) of the matrix is substituted into the Kalman filter prediction equation as the initial value to perform the second filtering to obtain the restored image. The simulation results show that the alternative Kalman filter restoration algorithm can quickly complete image restoration while removing degraded noise and ensuring image clarity. The calculation time is reduced to 2% of the calculation time of the Wiener filter restoration method.