一种基于参考图频谱相关的模糊参数判别方法

对于模糊图像的复原,模糊模型参数判别的准确性是一关键,针对大尺度模糊图像有效细节信息的丢失,噪声等干扰对模型特征的减弱,以及实拍图像的频谱特征不明显等问题,提出了一种基于参考图像频谱相关性的模糊参数判别方法,将一幅清晰图像按照不同模糊参数进行模糊处理,并与待复原图的频谱进行相关性分析计算,以获得准确的模型参数。实验结果证明,该方法能够适应大尺度散焦模糊,具有较强的抗噪性并且适应实拍的散焦模糊图像复原。

Blur parameter identification method based on spectrum correlation with reference image

The restoration of blurred image is highly dependent on accurate identification of blur parameters.Aiming at the loss of effective detail information in some rather blurred images,the decline of model features causing by noises,and the weakness of spectrum characteristics in photographed images,a method is proposed in this paper to identify the blur parameters based on spectrum correlation with reference image.A clear image chosen randomly is blurred with various parameters of the same model using this algorithm,and the spectrum correlation between the blurred image and the image to be restored is analyzed to obtain the right model parameters.Experiment results prove that the algorithm has anti-noise ability,and suits large scale defocused blurred images and the photographed defocused blurred images.