Depth-from-defocus: blur equalization technique

Abstract

A new spatial-domain Blur Equalization Technique (BET) is presented. BET is based on Depth-from-Defocus (DFD) technique. It relies on equalizing the blur or defocus of two different images recorded with different camera parameters. Also, BET facilitates modeling of images locally by higher order polynomials with lower series truncation errors. The accuracy of BET is further enhanced by discarding pixels with low Signal-to-Noise ratio by thresholding image Laplacians, and relying more on sharper of the two blurred images in estimating the blur parameters. BET is found to be superior to some of the best comparable DFD techniques in a large number of both simulation and actual experiments. Actual experiments used a large variety of objects including very low contrast digital camera test charts located at many different distances. In autofocusing experiments, BET gave an RMS error of 1.2% in lens position.

Depth from Defocus Technique Based on Cross Reblurring

“…If d is equal to the true depth D and the blurring with g z (i, j) in the discrete domain well simulates one in the continuous domain, the reblurred images f 01 and f 10 coincide, which was also noted in Ref. [18]. This property implies that the true depth D can be estimated by finding d in such a way that the corresponding g 1 and g 0 may render f 01 and f 10 to coincide each other.…”

“…To avoid the difficulty, the authors in Ref. [18] proposed a revised version of this approach based on "blur equalization" written as…”

“…2, Xian and Subbarao proposed a similar approach in Ref. [18] referred to as "blur equalization". They applied the property to the S-transform while the proposed technique directly formulates the property as in Eqs.…”

“…The proposed technique was compared with typical DFD techniques in Refs. [18], [19]. These techniques were first modified so that they can be applied to the synthetic and real test images in the previous subsections.…”

Exact Blur Measure Outperforms Conventional Learned Features for Depth Finding

“…These are called camera-based and image-based DFD equation pairs, with the solutions for the objective blur or depth dependent blur of both image points. The image based equation is obtained by local computing on the images based on an analysis in the frequency( [15] , [16] and [17]) or spatial domain ([ [18], [19] and [20]).…”