Automatic detection of residential buildings using LIDAR data and multispectral imagery

Presents two automatic building detection techniques using multispectral imagery and LIDAR data.

Main steps are:
1) Generate building masks: applies a height threshold to divide point cloud into ground and non-ground points. These two sets of points are used to generate two building masks namely primary and secondary building mask.
2) Line extraction: extracts lines around buildings from the primary building mask.
3) Form initial/candidate buildings and extend buildings: uses the extracted lines to form initial buildings and then (depending on the techniques) use NDVI and/or entropy from the multispectral imagery to extend candidate buildings. The secondary mask is also used during the extension.

4) Remove trees: a set of rules is applied to remove trees which are as high as buildings, also may have tree like shapes and colours.

For detail algorithms, please read
1. M. Awrangjeb, M. Ravanbakhsh and C. S. Fraser, “Automatic detection of residential buildings using LIDAR data and multispectral imagery,” ISPRS Journal of Photogrammetry and Remote Sensing, 65(5), 457-467, Sept. 2010.
2. M. Awrangjeb, C. Zhang and C. S. Fraser, "Building Detection in Complex Scenes Thorough Effective Separation of Buildings from Trees," Photogrammetric Engineering & Remote Sensing (PE&RS), vol. 78(7), 729-745, July 2012.