High-resolution image acquisition based on temporal integration with hierarchical estimation of image warping

As an approach towards improving spatial resolution of image acquisition, we have previously presented a prototype specific temporal integration imaging method which is tuned to a particular type of application where a user indicates a region of interest (ROI) on an observed image in advance and hence does not involve global image segmentation at all. The method uses a sub-pixel registration algorithm which describes image motion within the ROI with sub-pixel accuracy as deformation of quadrilateral patches covering the ROI and then performs a sub-pixel registration by warping an observed image with the warping functions recovered from the deformed quadrilateral patches. However, the size of quadrilateral patches depends on curvature of an object's surface, and hence it is extremely difficult to determine the proper size. To solve this problem, we introduce a hierarchical patch splitting algorithm for controlling spatial fineness of quadrilateral patches. The experimental simulations demonstrate that the temporal integration imaging method equipped with the capability of automatically controlling spatial fineness of quadrilateral patches works well for a smoothly curved surface the local geometrical structure of which can be well approximated as a rigid two-dimensional plane.