基于波前探测的视网膜图像半盲解卷积复原

获取高分辨视网膜图像的难点在于是否能够消除成像系统中像差的影响。为了进一步改善图像质量、提高视网膜细胞组织的观察分辨率,提出了结合自适应光学成像技术和视网膜图像后处理算法的方法。通过视网膜成像系统中的自适应光学技术实时校正人眼像差并获取初始视网膜图像,根据成像系统中的残余像差重建光学传递函数作为图像复原模型初始参数估计,最后对视网膜图像进行条件约束迭代半盲解卷积复原,进一步消除像差对成像质量的影响,从而得到高分辨率视网膜图像。实验结果表明:由这种方法处理的视网膜图像质量能得到明显提高,其图像质量客观评价参数(GMG,LS和PSV)比原始图像提高近1倍,在视网膜细胞的空间频率范围内(70～90 cyc/deg),复原后图像的功率谱平均值比原始图像提高了10倍左右,基本能满足观察分辨率要求。

Retinal Image Semi-Blind Deconvolution Restoration Based on Wave-Front Detection

The difficulty to obtain high resolution retinal image lies in eliminating the effect of aberration which exists in the imaging system.The proposed combination of adaptive optics imaging technology and retinal image post processing algorithm can improve image quality and observation resolution.The aberration of the human eye can be corrected in real time by adaptive optics technology in retinal imaging system and the original retinal image can be obtained,and the optical transfer function which is used as initial parameter estimate of image deconvolution modal can be constructed with the residual aberration of the imaging system.Finally,iterative semi-blind constraint deconvolution restoration is carried out on the retinal image,so as to eliminate the influence of residual aberration on imaging quality and obtain the high resolution retinal images.The experimental results show that the retinal image quality is improved significantly by the proposed method,the image quality objective evaluation parameters(GMG,LS and PSV) are increased by nearly one time over the original image,and the averaged power spectrum is increased by about 10 times in the spatial frequency range(70—90 cyc/deg) of retinal cells,all of which can meet the requirement of observation resolution.