Design of a three-dimensional near real-time image processing system

Abstract

The airborne three-dimensional (3D) imager, which integrates global positioning system (GPS), attitude measurement unit (AMU), scanning laser rangefinder (SLR) and spectral scanner, has been developed successfully. The spectral scanner and SLR use the same optical system, which ensures that the laser point matches pixel seamlessly. The distinctive advantage of the airborne 3D imager is that it can produce geo-referenced imagery and DSM (digital surface models) without any ground control points (GCPs). It is no longer necessary to obtain GCPs, and the software can process the data to produce DSM and a geo-referenced image in near real-time, so the airborne 3D imager is many times faster than the traditional approaches. The processing procedure of the software involves decomposing and checking the raw data, processing GPS data, calculating the positions of laser sample points, producing a geo-referenced image, producing DSM, and mosaicking strips. In this paper, we introduce the principle of the airborne 3D imager first and then focus on the fast processing algorithm. The flight tests and processed results demonstrate that the processing technique is feasible, and can meet the requirement of near real-time applications.