Article citationsMore >>

Arefi, H., Engels, J., Hahn, M. and Mayer, H. (2008). “Levels of detail in 3d building reconstruction from lidar data”. International archives of the photogrammetry, Remote sensing and spatial information sciences. vol. xxxvii, part b3b., pp. 485-190.

has been cited by the following article:

Article

LIDAR Image Segmentation Using Hierarchical Clustering

1Islamic Azad University Fasa Branch

2Shahid Chamran University of Ahvaz


American Journal of Computing Research Repository. 2016, Vol. 4 No. 1, 21-29
DOI: 10.12691/ajcrr-4-1-4
Copyright © 2016 Science and Education Publishing

Cite this paper:
Tahereh Sharafi, Gholamreza Akbarizadeh. LIDAR Image Segmentation Using Hierarchical Clustering. American Journal of Computing Research Repository. 2016; 4(1):21-29. doi: 10.12691/ajcrr-4-1-4.

Correspondence to: Tahereh  Sharafi, Islamic Azad University Fasa Branch. Email: tahereh.sharafi@ymail.com

Abstract

Hierarchical clustering method is adopted for LIDAR image segmentation after extracting the intended features for identifying complex objects. In the experiments, four LIDAR images with different numbers of areas (sea, forest, desert, and urban) were used for examining the algorithm. The efficiency of image segmentation was generally evaluated visually because the segments of the main image typically lack certain, fixed features and depend on the criterion used for pattern distance/similarity as well as the threshold for cluster separation. For each experiment, hierarchical clustering method was employed by creating the clustering hierarchy tree and specifying the optimal number of clusters on the basis of tree data. Once the optimal number of clusters was determined, the similarity matrix of data image patterns was separated according to Euclidean distance algorithm in terms of greatest similarity among the patterns to the number of clusters. Clustering was then performed. The program output comprised labeled images for samples specifying which pattern pertains to which cluster. The images associated with each cluster are displayed separated from other clusters with other areas eliminated. The results indicated that for LIDAR images that lack a certain, fixed feature, the hierarchical clustering method for segmentation can perform separation and labeling.

Keywords