Depth Estimation from a Single Defocused Image Using No Reference Image Quality Assessment Metrics

In this paper, a new algorithm is proposed for estimating depth from defocus. Most of the proposed methods use at least two blurred images to estimate the relative depth of objects lying in a scene. The proposed method uses just one image to do that. It is based on image quality assessment criteria. A deblurred image can be divided into small patches. Each patch is deblurred with various scales of PSF. A deblurred result which yields the best quality is used to determine the correct PSF scale. The quality is measured with a weighted sum of four image quality assessment measures. Extensive simulations are conducted to compare the performance of our method with the previously proposed ones.

Keywords:

Depth estimation , Image quality assessment , Depth from defocus.