To use a projector anytime anywhere, a lot of projector-camera based approaches have been proposed. There are mainly three categories in these projector-camera based approaches. Geometric warping is for correcting geometric distortion or alignment of multiple projections. Radiometric compensation corrects color variation caused by color of a screen object or environmental lighting. Focal correction reduces the effect of defocus blur which occurs when a screen is located out of projector's depth-of-field. To achieve focal correction, we have to estimate the degree of the defocus blur on the displayed image. Defocus blur on the displayed image is represented as a Point-Spread-Function (PSF). There are mainly two methods to estimate PSF; one is marker pattern projection based method, another is based on evaluation of degree of defocus blur of projected image. However, both of them estimate PSFs at every sub-image that is divided pre-determined geometry. The accuracy of PSF estimation depends on the content of projection image. When the projection image is less-textured or already blurred, the estimated PSF is not accurate. Because less-textured or already blurred region is little-affected by projector defocus blur. In this paper, we propose a novel method for estimating PSF without projecting any special pattern images. To prevent PSF estimation error, we extract well-suited region to estimate PSF in the projection image and estimate PSF at extracted region. Experimental results show that our method can estimate the PSF more accurate than our previous method, in which fixed region and sub-image is used for PSF estimation.