A New Sense for Depth of Field
IEEE Transactions on Pattern Analysis and Machine Intelligence
Depth from defocus: a spatial domain approach
International Journal of Computer Vision
Depth from defocus estimation in spatial domain
Computer Vision and Image Understanding
3-D Shape Estimation and Image Restoration: Exploiting Defocus and Motion-Blur
3-D Shape Estimation and Image Restoration: Exploiting Defocus and Motion-Blur
Hi-index | 0.00 |
The method proposed in this paper uses two blurred images, acquired from the same point of view with different focus settings, in order to estimate depth in real time. It falls under the group "Depth from defocus" methods. The blur is modelled by the convolution of a Gaussian Point Spread Function (PSF) which with the theoretical sharp image. We calculate the gradients and Laplacians of the two blurred images and according to the type of contour, step, ramp, roof o line, we compute the differences of these images or their derivative until order 2 to obtain the difference in blur, difference of the variances of the Gaussian ones, which we connect to the depth by taking account of the parameters of the optical system. We use then this difference to the depth by tacking account of the parameters of the optical system. We present a set of results on real images showing the performance of the method and its limits. The computing times which we measured make possible the use of our algorithm at video rate.