相关序列图像复原算法研究

针对气动光学效应降晰过程的复杂性,提出了一种基于图像相关性的两帧序列图像复原算法。该算法利用图像的相关性,结合气动光学效应降晰函数辨识,在大气流场的光学退化模型未知的情况下,采用两帧序列图像降晰函数辨识实现气动光学效应退化图像复原。为了克服降晰函数辨识算法可能带来的误差,在序列图像辨识降晰函数时,提出将辨识得到的两帧序列图像降晰函数结果差的二阶范数作为惩罚项,当该范数不符合要求时,根据序列图像的质心偏移和旋转偏移对图像进行相应的修正。给出了实现该方法的具体步骤,仿真结果证明了算法的有效性。

Restoration algorithm of sequence images in aero-optics

To solve the complexity of blurring in aero-optics,a method of restoration based on correlation of two series images is proposed.Using correlation of images and combining identification of blurs in aero-optics effects,degraded images can be restored by use of two series images with identification of blurs under the condition of the degraded models being unknown.In order to overcome the error caused by identification of blurs,norm of the difference of PDFs identified with two series images is considered as the penalty terms and when the norm is not accord with the request,images are revised according to centroid and angle motion.The proposed method is studied and the experimental results illustrate the efficiency of the method.