The interference fringes obtained by optical methods such as holographic interterometry and Twyman-Green interferometry are analyzed using image processing. To perform accurate and automated analysis, phase analysis methods have been recently developed. The Fourer transform moire and grid methods (FTMGM) and the phase-stepping method (PSM) are popular techniques to analyze phase distribution. However the FTMGM has noticeable error at the edges of the image if the data at both edges are not continuous. The PSM has large error if the brightness has noise. A new phase analysis method is proposed to erase these errors. This method can be performed using 3-D image processing. A 3-D (x,y,α) image is formed of many sequential phase-shifted 2-D (x,y) images, where a is the shifted phase value. By performing the a-directional Fourier transform on this 3-D image, the accurate phase distribution of the fringe pattern is obtained. An example applied to Twyman-Green interferometry is shown.