When machine vision technology is used to detect the inner surface quality of a pipeline, the geometric center of an original image will deviate the center of a pipe wall panoramic annular image in the original ones fore ground area. However, it will lead to serious deviation in the image unwrapped with the geometric center. In order to reduce the measuring errors of the geometry characters, a fast unwrapping algorithm for the panoramic annular image with center deviation was proposed. Firstly, the least square fitting circle method was used to find a fitting circle representing a cross-section of the wall on the image and to obtain the center parameters of panoramic annular image. Then, by unwrapping the panoramic annular image with a quarter circle scanning method and by correcting the frame aspect ratio of the unwrapping image, the panoramic image was completely restored. Experimental results indicate that it takes 0.868 s to unwrap the annular detection area of a panoramic image into a rectangular one with the resolution of 2 044 pixel ×199 pixel, and the relative errors in outline dimensions of the tested targets are less than 1.54%. It is shown that the method could satisfy the real time and accuracy requirements in the image processing.