Many image processing applications require a quantitative estimation of the underlying correspondence (of the relative position in space) between two or more images. Here, we propose a unique, less numerically intensive method, which we refer to as the subtraction method (SM), to measure in-plane object motion with subpixel accuracy. This method can be usefully employed in conjunction with the correlation method (CM) to provide fine motion information very quickly (∼100 times faster). We first review the CM, then, demonstrate, explain, and examine the proposed SM using analytic functions, standard digital images, and using both simulated and experimental data. Specifically, subpixel motion results for Gaussian functions, four different standard images, and speckle images are examined. Performance is compared to that of the CM. Experimentally, images are captured using both optical and THz imaging systems. A high-resolution megapixel camera is used in the optical system and a low-resolution camera with 16 × 16 pixels is used to perform the THz measurements. One-dimensional subpixel motion in both x and y is examined and analyzed. The results demonstrate that the method can be flexibly used in different areas.
Read full abstract