Optical Ortho-Rectification for Image-Based Stream Surface Flow Observations Using a Ground Camera

Abstract

Image-based stream flow observation consists of three components: (i) image acquisition, (ii) ortho-rectification, and (iii) an image-based velocity estimation. Ortho-rectification is a type of coordinate transformation. When ortho-rectifying a raster image, pixel interpolation is needed and causes the degradation of image resolution, especially in areas located far from the camera and in the direction parallel to the viewing angle. When measuring the water surface flow of rivers with a wide channel width, reduced and distorted image resolution limits the applicability of image-based flow observations using terrestrial image acquisition. Here, we propose a new approach for ortho-rectification using an optical system. We employed an optical system embedded in an ultra-short throw projector. In the proposed approach, ortho-rectified images were obtained during the image acquisition phase, and the image resolution of recorded images was almost uniform in terms of physical coordinates. By conducting field measurements, characteristics of the proposed approach were validated and compared to a conventional approach.