In order to solve the problem of blurred license plate image in road monitoring, the PSF estimation method based on spectrum is often used to restore the blurred license plate image. Due to the large error of motion blur information estimation, a blind license plate restoration algorithm based on image spectrum and cepstrum is proposed in this paper. First, Radon transform is used to detect the parallel line features in the spectrum of fuzzy license plate image to estimate the fuzzy angle. Then, the cepstrum data of the blurred image is used to measure the peak value and estimate the fuzzy length of the image. Finally, blind restoration of fuzzy license plate image is carried out by Wiener filter of autocorrelation function. The experimental results show that the PSF estimation result of this algorithm is more accurate, and it can achieve better restoration effect of fuzzy image for license plate images under different fuzzy conditions. It is verified that the blind restoration algorithm based on fusion of image spectrum and cepstrum has certain feasibility and adaptability.

中文翻译：

---

基于频域特征的移动模糊车牌图像盲恢复算法新方法及实现

为了解决道路监控中车牌图像模糊的问题，常采用基于谱的PSF估计方法来恢复模糊的车牌图像。针对运动模糊信息估计误差较大的问题，提出一种基于图像频谱和倒谱的盲车牌恢复算法。首先，利用Radon变换检测模糊车牌图像谱中的平行线特征来估计模糊角度。然后，利用模糊图像的倒谱数据测量峰值，估计图像的模糊长度。最后，利用自相关函数的维纳滤波器对模糊车牌图像进行盲复原。实验结果表明，该算法的PSF估计结果更加准确，对不同模糊条件下的车牌图像能达到较好的模糊图像复原效果。验证了基于图像频谱和倒谱融合的盲恢复算法具有一定的可行性和适应性。