Abstract:
Several methods have been proposed to correct motion in medical and non-medical applications, such as optical flow measurements, particle filter tracking, and image registration. In this paper, we designed experiments to test the accuracy and robustness of a recently proposed algorithm for subpixel image registration. In this case, the algorithm is used to correct the relative motion of the object and camera in pairs of images. This recent algorithm (named phase-based Savitzky-Golay gradient-correlation (P-SG-GC)) can achieve very high accuracies in finding synthetically applied translational shifts. Experiments were performed using a camera, a flat object, a manual translational stage, and a manual rotational stage. The P-SG-GC algorithm was used to detect the flat object motion from the initial and shifted images for a set of control points on the surface of the object, which were automatically matched in subimages of 128 pixel × 128 pixel. A least-squares method was used to estimate the image transformation matrix that can register the shifted image to the initial image. The results demonstrated that the P-SG-GC algorithm can accurately correct for the relative motion of the object and camera for a large range of applied shifts with a registration error less than 1 pixel. Furthermore, the P-SG-GC algorithm could detect the images in which the motion could not be corrected due to poorly matched control points between the initial and shifted images. We conclude that the P-SG-GC algorithm is an accurate and reliable algorithm that can be used to correct for object or camera motion.