中文摘要：

为找到一种基于匀速直线运动模糊的单帧图像，准确辨识成像点扩散函数中模糊参数的方法，对现有文献提出的倒谱、Radon变换、图像微分、图像自相关性以及检测函数等算法进行了仿真和比较。仿真结果表明，倒频谱算法能够更准确地辨识精度运动方向；像素相关性方法能够更准确地辨识模糊长度。对匀速直线运动模糊的单帧图像进行模糊参数辨识时，结合倒谱辨识运动方向和像素相关性辨识模糊长度的算法，能得到更高精度的辨识结果，利于更好地复原图像的质量。

英文摘要：

To search a parameter identification method with high accuracy for single image blurred by uniform linear motion, a few of available algorithms, such as cepstral analysis, Radon transform, image derivative, autocorrelation-based and detect function-based, were used to simulate the performance of identification and to compare the results of parameter identification, especially for parameters of motion direction and blur length. It was found that cepstral analysis algorithm could identify motion direction with the best accuracy, while autocorrelation-based method could identify blur length with the best accuracy. Based on that analysis, it could be suggested that the scheme combined of cepstral analysis with autocorrelation-based algorithms would get high accurate estimation when single image is just blurred by uniform linear motion.