Out-of-focus blur estimation and restoration for digital auto-focusing system

A fully digital auto-focusing system based on point-spread function (PSF) estimation is proposed. To increase accuracy in estimating the PSF of the defocused image, the input image is subdivided, and the step responses of edge-containing subimages are averaged. The PSF, which results from out-of-focus blur, is estimated by differentiating the averaged step response along the appropriately estimated edge direction under the assumption that an image usually contains a piece-wise linear boundary between an object and background. By using the estimated PSF, the corresponding image restoration filter is realised, based on the constrained least squares approach