Improving 2D change detection by using available 3D data

Improving 2D change detection by using available 3D data

Van der Sande, C.J.
Zanoni, M.
Gorte, B.G.H.

Aerospace Engineering

Remote Sensing

2008-07-01

Change detection with very high resolution imagery is difficult, because 3D objects as buildings appear differently in 2D imagery due to varying viewing angles and sun positions. This research proposes a method to improve change detection by using simple 3D models of buildings. Buildings have been modelled as flat roof or gable roof buildings. By knowing the position of the sun, the location of the shadow of a building is calculated. The location of the relief displacements is calculated using information on the position of the sensor. By introducing the projected shadow, relief displacement and roof in the automatic change detection process more reliable change signals are produced. An object-oriented approach for change detection is developed using segmentation techniques to create homogeneous segments from pixels. The method compares the object map t0 with the image t1. Information is gathered on how each object represents in the image t1 and at t0, by using the information of the image t0. In the case of shadow parts histogram stretched imagery are used. For the change detection on the roof, the roof segments are reprojected on top of each other to compare them. This method resulted in more reliable change detection results and increased the detection rate to 72% for changes to buildings.

remote sensing
change detection
updating
segmentation
3D
object oriented

http://resolver.tudelft.nl/uuid:b92f2beb-5dac-47c8-84ab-f5e681c8081c

International Society for Photogrammetry and Remote Sensing

XXI ISPRS Congress, Commission I-VIII, 3-11 July 2008, Beijing, China

Institutional Repository

conference paper

(c) 2008 The Author(s)