基于边缘细化的散焦失真图像恢复方法

Defocus distortion image restoration method based on edge thinning

为避免单幅散焦图像中光栅及空间不确定性的影响,改善图像的峰值信噪比、信息熵与平均梯度,提出基于模糊聚类的单幅散焦图像深度恢复方法。以原单幅散焦图像为依据,采用多分辨率分析法规划逐步下降分辨率的图像簇,并通过无监督模糊聚类法将物体轮廓化,完成单幅散焦图像物体与背景的分割。采用边缘细化暗通道去雾算法调整分割得到的物体轮廓边缘景深跳变位置的暗通道值,降低晕光现象。通过添加修正因子,减轻图像背景区域色彩失真情况,优化透射率,完成单幅散焦图像深度恢复。经实验验证:该方法恢复后的图像具有较高平均梯度、信息熵与峰值信噪比,且运算时间低,恢复后图像可见边数目多,说明该方法的图像恢复效果较好。

In order to avoid the influence of grating and spatial uncertainty in a single defocused image and improve the peak signal-to-noise ratio,information entropy and average gradient of the image,a depth restoration method of a single defocused image based on fuzzy clustering is proposed in this paper.According to the original single defocus image,the multi-resolution analysis method was introduced to gradually reduce the resolution of the image cluster.The object was contoured in order to segment the object and background of a single defocus image via unsupervised fuzzy clustering.The edge thinning dark channel defogging algorithm was applied to adjust the dark channel value of the depth of field jump position at the edge of the segmented object contour,thus reducing the halo phenomenon.The correction factor was added to reduce the color distortion of the image background area,so as to optimize the transmittance and complete the depth recovery of a single defocus image.The experimental results show that the restored image has high average gradient,information entropy and peak signal-to-noise ratio,low operation time,more visible edges and excellent image restoration effect.