一种基于扩散驱动先验技术的图像超分辨率重构方法

An Image Super-resolution Reconstruction Method Based on Diffusion Driven Prior

当输入图像因污迹、噪声和采样而严重退化时,目前基于Papoulis-Gerchberg(PG)算法的大多数超分辨率方法表现不佳.因此,提出了一种基于扩散驱动先验和PG算法的超分辨率方法,能够在提高图像分辨率的同时,估计缺失的高频分量.首先提出了一种新型扩散驱动平滑的先验,能够在平坦和轮廓区域之间自动平衡作用,确保正则化水平以产生清晰图像.然后,将PG算法引入到迭代过程中,以估计重构场景中缺失的小规模特征.实验结果表明,相比现有的超分辨率方法,提出方法的峰值信噪比和结构相似指数结果更高,重构图像更加清晰且无伪影.

Most of the super-resolution methods currently based on the Papoulis-Gerchberg(PG) algorithm perform poorly when input images are heavily degraded by smudge,noise and sampling.Therefore, a super-resolution method based on diffusion driven priors and PG algorithms is proposed to estimate the missing high-frequency components while improving image resolution.First,a novel diffusion driven smoothing priors that can automatically balance between flat and contour regions are developed to ensure regularization levels to produce clearer images.Then,the PG algorithm is introduced into the iterative process to estimate the small scale features in the reconstructed scene.The experimental results show that compared with the existing super-resolution methods,the proposed method has higher peak signal-to-noise ratio(PSNR) and structure similarity index(SSIM),and the reconstructed image is clearer and without artifacts.