Potential and idiosyncrasy of Object‐based image analysis for airborne Lidar‐based building detection (CROSBI ID 249318)
Prilog u časopisu | izvorni znanstveni rad | međunarodna recenzija
Podaci o odgovornosti
Tomljenović, Ivan ; Blaschke, Thomas ; Hofle, Bernhard ; Tiede, Dirk
engleski
Potential and idiosyncrasy of Object‐based image analysis for airborne Lidar‐based building detection
Today, a plethora of approaches exists that can be used for the task of object extraction from LiDAR data (point and/or derived raster information). The majority of approaches deal with the extraction of artificial objects or distinct natural objects (e.g. single trees). Local height variations or number of signal returns are often input to extraction processes. If only relying on pixel‐based concepts, results of such analyses may provide sufficient detection accuracy but they may also generate various misclassified objects since they lack object‐based view. In the presented paper, we developed an Object‐Based Image Analysis (OBIA) approach for building extraction starting from LiDAR point data only. We hypothesize that an OBIA approach could performs better than a pixel‐based approach as it resembles the way in which humans perceive reality. The developed method consists of a set of rules applied to continuously re‐shaped (based on pre‐classifications) candidate objects. The results of the proposed approach have been compared to buildings which were manually identified from a Digital Surface Model (DSM). Primary results show completeness value of 97.80% and correctness of 80.5%.
Remote Sensing ; Geography ; LiDAR ; Image processing and analysis
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
Podaci o izdanju
3 (2S)
2014.
517-520
objavljeno
2241-1224