Semiglobal Matching Results on the Isprs Stereo Matching Benchmark

Abstract

ABSTRACT:Digital surface models can be efficiently generated with automatic image matching from optical stereo images. The Working Group 4 of Commission I on "Geometric and Radiometric Modelling of Optical Spaceborne Sensors" provides a matching benchmark dataset with several stereo data sets from high and very high resolution space borne stereo sensors at http://www.commission1.isprs.org/wg4/. The selected regions are in Catalonia, Spain, and include three test areas, covering city areas, rural areas and forests in flat and medium undulated terrain as well as steep mountainous terrain. In this paper, digital surface models (DSM) are derived from the Cartosat-1 and Worldview-1 datasets using Semiglobal Matching. The resulting DSM are evaluated against the first pulse returns of the LIDAR reference dataset provided by the Institut Cartogràfic de Catalunya (ICC), using robust accuracy measures.

A Detailed Study About Digital Surface Model Generation Using High Resolution Satellite Stereo Imagery

“…Its biggest benefit is that the GSD of generated 3D point cloud is the same as the original stereo images (Haala, 2013). Recently, researchers have investigated and verified the availability and reliability of the dense stereo matching approach with VHR satellite imagery (Reinartz, et al, 2010;d'Angelo and Reinartz, 2011;Wohlfeil, et al, 2012). This paper applies the C/C++ library LibTsgm to implement the dense image matching and deliver the disparity images.…”

Detección de cambio en zonas urbanas utilizando MDTs generados automáticamente a partir de imágenes satelitales estereoscópicas de alta resolución espacial

“…The basic principle of this algorithm lies in a small processing window called kernel and a string of bits where each bit is assigned to a pixel within this window, taking the pixel a value of 1 if it has lower intensity than the central pixel. The main advantage of this transformation is its invariability to changes in the digital number, being very suitable for matching the stereo images with radiometric differences [26].…”

“…Techniques based on mutual information (MI) are more stable and they have the best performance according to Gehrke et al [25], since they are not sensible to changes in lightemission and recording techniques [24]. The calculation of costs based on Mutual Information is defined as the value of similarity for potential pixels to be matched in two images [26], called entropy by Hirschmüller [24] and requiring an epipolar geometry for them. That is when the scanning lines of the stereo pair are epipolar lines, which occurs when two axes of the stereoscopic camera system are parallel to each other and perpendicular to the base.…”

Automatic UAV Image Geo-Registration by Matching UAV Images to Georeferenced Image Data

“…The UAV images were acquired with a Sony Nex-7 camera simultaneously with the reference aerial images on 2 November 2015. For both scenarios, we matched UAV images not only to aerial images, but also to aerial orthophotos, which are generated by an orthographic projection onto a high resolution DSM [37,38]. The Germering dataset is comprised of four different scenarios: Container, Highway, Pool1 and Pool2.…”

“…If image R is an individual georeferenced aerial or satellite image, we assume its height map is available, which can be generated in the process of dense matching with neighboring images [37]. The height Z can be looked up in the height map, and the planar coordinates X and Y can be calculated using the orientation parameters of R. If image R is an aerial orthophoto that is generated by an orthographic projection of the aerial image mosaic onto a high resolution DSM, the planar coordinates (X, Y) are namely the corresponding georeferenced coordinates of the pixel (x r , y r ) in the orthophoto, and Z is namely the corresponding height at (X, Y) of the DSM.…”