【摘要】 在散焦图像中,点的模糊程度随物体的深度而变化,因此可以利用散焦图像中点的散焦程度来估计物体的深度信息。本文提出了一种基于散焦图像中物体的边缘梯度关系来恢复图像深度图的新算法,用一个已知参数的高斯函数对图像进行再模糊,然后求出模糊后的物体边缘梯度,再与原图像中物体边缘梯度相比,再将该比值与图像的深度关联,求出图像中物体边缘处的深度,再利用后续深度插值方法和深度图优化恢复出整幅图的深度信息。这种算法仅需要一幅图像即可进行深度信息恢复,有较好的有效性。更多还原

【Abstract】 In a defocus image,the blur of a defocus point varies with its depth,therefore,we can use the defocus information to estimate the depth map of a defocus image.In this paper,we propose a method which used the edge gradient of the objects in an image to recover depth map.We use a Gaussian function which its kernel is known to reblur the original defocus image and then we calculate the ratio between the original gradient and the reblurred edge gradient,this ratio can be related with the depth.So,the edge depth can be recovered.The depth map of the whole image can be recovered by a depth propagation method.This method only needs one defocus image and continuity is preserved.更多还原