| Abstract: | Segmentation followed by extraction of an object for LiDar (Light detection and ranging) data is an essential step in many applications, namely, hazard mapping, rock fall, data classification. This paper presents an approach for object segmentation from dense LiDar point cloud data. The concept of the proposed technique is based on statistical analysis over the derivative point cloud. Initially the data are divided into background and non-ground regions. Then, the non-ground area is further segmented into distinct objects using statistical information. Experimental results in a hilly area show that the proposed algorithm can efficiently segregate the rock unit. Comparing with other techniques, it offers the highest rate of correctness, fullness and rapidness for rock unit separation.
|