Error Analysis and Compensation in Images Stitching for the Mechanically Stitched CCD Aerial Cameras

Abstract

Image misalignment during image stitching is a common issue for the mechanically stitched CCDs in the aerial cameras due to the relative image motion between CCDs. In this paper, we analyze the error in imaging stitching for the mechanically stitched CCDs and propose a compensation method based on position and orientation system (POS). First, the imaging relationship of overlapping pixels in mechanical stitching is analyzed. The imaging model of adjacent CCD is constructed according to the collinear equation. The effects of stitching staggered distance, carrier attitude change and flight speed on the image mosaic of overlapping areas are given. Monte Carlo algorithm is used to analyze the statistical value of image mosaic error in overlapping area under typical working conditions. Then, a geometric correction method based on POS recording of external orientation elements of adjacent CCD imaging is proposed. The overlapping area stitching error is reduced from 14.9 pixels to 0.4 pixels which meets the engineering requirements. The mechanical stitching mathematical model, analysis and correction method established in this paper have strong engineering and application significance for mechanical stitching of aerial cameras.