3D Building Change Detection from High Resolution Spaceborne Stereo Imagery

Abstract

In this paper, a novel approach for 3D building change detection is proposed using Digital Surface Model (DSM) generated from High spatial Resolution Spaceborne Stereo (HRSS) imagery. To improve the change detection performance, the difference image is denoised by the detected shadow mask and DSM hole mask. Several thresholding algorithms are compared to remove spurious change in altitude caused mainly by computation errors in the DSM generation procedure. After applying the thresholding methods, object-oriented image analysis approach is carried out by comparing the size and shape properties of the objects that are extracted from the thresholded difference image to filter small changed regions. The performance evaluation of the proposed 3D building change detection confirms the effectiveness of this approach.

Building Change Detection Based on Satellite Stereo Imagery and Digital Surface Models

“…Another important effect is shadowing which can significantly influence the quality of the DSMs. It has been shown in our previous paper [17] that shadow areas usually result in relatively bad matching results. Matching failures in shadow areas, which often represent ground level, are displayed in the original generated DSMs partly as holes, and through interpolation methods, they often get higher height values than the ground level.…”

“…After generating the building change mask, it is still required to separate "changed building" from false change alarms. We therefore apply an edge-based building extraction method and improve the output by extracting the undesired objects based on their shape properties [17]. The three most important features to differentiate building areas from other objects are height, area (size), and convexity.…”

Region Based Forest Change Detection From Cartosat-1 Stereo Imagery

“…First, with 2.5 meter resolution Cartosat-1 imagery single trees can not be separated and thus tree detection based on shape features (Mora, 2010;Tian, 2011) can not be adopted for this application. Second, many land cover classes around the forest have similar gray values, such as rivers, shadows and also manmade constructions.…”

Multitemporal 3D Change Detection in Urban Areas Using Stereo Information from Different Sensors

“…This is partly due to better availability, lower cost, higher time resolution and larger coverage ability. With the height difference images, the changes can already easily be seen from the images [8][9]. But how to locate the real changes and how to highlight the exact size of the changed area in 3 dimensions, especially from satellite data is still a challenge that needs to be solved.…”

“…All of this will influence the quality of the generated DEMs, which will finally lead to a lot of artificial changes in the extracted change map. In former studies, we tried to classify the real change by analyzing the changed region in the difference image by using shape features [9]. That method works well in commercial and suburban areas, which exhibit large buildings and houses of similar shape, while for the centres of big cities, which have more complicated building structures, it is quite difficult to define a common character of the buildings.…”