Accurate 3-D shape recovery using curved window focus measure

Abstract

To improve the accuracy of 3-D shape recovery from image focus, a new shape from focus (SFF) method based on focused image surface (FIS) is proposed. The new SFF method, called the Curved Window Focus Measure (CWFM), uses piecewise curved search windows which track the realistic FIS in image space to compute the focus measure. The Nine Control Points (NCP) model is proposed to interpolate the piecewise curved window, and the matrix form of the Lagrange polynomial is derived as a surface equations appropriate for the NCP model. An automated CCD camera system has been developed to implement the CWFM search method. Experimental results show that our CWFM search method gives more accurate results than the previous SFF methods.