基于点扩散函数支持域扩展的多帧未匹配退化图像盲解卷积复原

Multi-frame unregistration degraded image blind deconvolution based on point spread functions' support domain expansion

多帧盲解卷积算法利用多帧退化图像进行复原可以获得清晰原始图像和点扩散函数信息,受到了很多研究学者的关注。目前大部分多帧盲解卷积算法都需要对多帧退化图像进行匹配预处理,消除图像间平移引入的算法求解误差。利用频率内的多帧盲解卷积算法对未匹配的退化图像进行处理,不需要进行预匹配处理,只需要对点扩散函数的支持域进行扩展就可以复原获取清晰化的图像。利用傅里叶变换性质对该方法的可行性进行了说明。同时对该方法进行了数字仿真实验,复原结果中的点扩散函数发生相对移动消除了图像间未匹配的影响,证实了本文方法的有效性。

Multi-frame blind deconvolution algorithm using multi-frame degraded images can get the information of clear original image and point spread functions,which was concerned by many researchers.At present,the majority of multi-frame blind deconvolution algorithms needed registration pretreatment for multi-frame images to eliminate the restored image error.Multi-frame blind deconvolution algorithm in frequency domain was proposed to process the degraded images without registration and though expanding point spread functions＇ support domain.The feasibility of this method was illustrated by the properties of the Fourier transform.The digital simulation experiments were carried out to confirm the effectiveness of the proposed method.