Analysis of Epipolar Geometry in Linear Array Scanner Scenes

Abstract

Resampled imagery according to epipolar geometry, usually denoted as normalised imagery, is characterised by having conjugate points along the same row (or column). Such a characteristic makes normalised imagery an important prerequisite for many photogrammetric activities such as image matching, automatic aerial triangulation, automatic digital elevation model and orthophoto generation, and stereo viewing. The normalisation process requires having the input imagery in a digital format, which can be obtained by scanning analogue photographs or by direct use of digital cameras. To reduce the time gap between the data acquisition and product delivery, many small‐scale mapping projects now rely on digital cameras. Digital frame cameras still, in general, provide imagery with geometric resolution and ground coverage inferior to scanned images from analogue cameras. Linear array scanners (line cameras) have therefore emerged as a possible alternative to digital frame cameras, especially for high‐resolution space‐borne imaging, with performance comparable to that of analogue frame cameras. The normalisation process of frame images is a well‐established and straight forward procedure. On the other hand, the normalisation process of linear array scanner scenes is not as straightforward and is sometimes mysterious. For example, providers of space‐borne imagery furnish normalised line scanner imagery while the user community is not aware of the underlying process. This paper presents a comprehensive analysis of the epipolar geometry in linear array scanner scenes. Special emphasis is directed towards scanners moving with constant velocity and attitude since such a trajectory closely resembles the imaging geometry of the majority of current space‐borne scanners. The research presented highlights the factors affecting the shape of the resulting epipolar lines such as the stereo coverage configuration and the geometric specifications of the imaging system. In addition, the paper outlines a comparative analysis of the normalisation process for frame and line cameras. The presented concepts are verified through experimental results with synthetic data.

Unified piecewise epipolar resampling method for pushbroom satellite images

“…It was based on the researches presented by Habib et al [7] and Kim [2]. These previous works showed that epipolar curves of linear pushbroom images are not lines but hyperbola-like non-linear curves and that an epipolar curve for the entire image can only be approximated by piecewise linear segments, with epipolar curve pairs Fig.…”

Satellite Stereo Based Digital Surface Model Generation Using Semi Global Matching in Object and Image Space

“…In contrast to frame cameras, the epipolar geometry of the satellite pushbroom sensors shows different behaviour. The epipolar curves of the pushbroom cameras are not straight and the conjugate pair does not exist (Morgan et al, 2004;Oh et al, 2010;Habib et al, 2005). Therefore, the commonly used techniques for stereo rectification of frame cameras can't be directly applied for rectification of images from pushbroom sensors.…”

“…Several techniques have been proposed for pushbroom rectification with the affine camera model approximation being the most popular solution for pushbroom rectification. The affine assumption is suitable due to the following reasons: the height variations in the object space are small compared to the satellite altitude, the angular field of view is small and the satellite is assumed to travel in a straight line with a constant velocity (Morgan et al, 2004;Habib et al, 2005) during the short image acquisition time. An affine transformation is estimated using corresponding points between the two images and the resulting transformation causes the corresponding points to lie along the same image rows.…”

An automatic and modular stereo pipeline for pushbroom images

“…Using various pushbroom camera models (Orun andNatarajan, 1994, Hartley andGupta, 1997), it can be shown that pushbroom image pairs have non-straight epipolar curves and that these curves are not conjugate, making rectification impossible (Kim, 2000, Habib et al, 2005.…”