Feature Extraction (P2.02)

This research uses both airborne imagery and LiDAR point clouds for feature extraction. The first significant deliverable will be a process to metrically combine the imagery and LiDAR point clouds to form accurate 3D images. The combined 3D images provide additional information through each of the data sources. These 3D images will be used to automatically extract features for modelling objects such as buildings and vegetation parameters. Further computational tools are being developed to operate and process multi-sensor datasets.

With digital imaging and laser ranging technology advancing over the last decade, both high-resolution commercial satellite imagery, and imagery and laser scan data captured from digital aerial sensors have provided new data sources for spatial information generation.

These data display enhanced spatial, spectral and temporal resolution, allowing for accurate and reliable detection and characterisation of the changing earth in ever more detail.

There is an ongoing explosion in the amount of spatial data provided by new digital sensors. However the associated production of spatial information products is constrained by the slow, expensive and manually intensive processes of feature extraction for mapping and GIS.

Highlights

The registration of imagery and 3D point clouds has been improved through applying variable intervals to classify the representation of the data. This has reduced the failure rate by 10%.

The methodology developed for extraction of building footprints has been tested on larger datasets. The reconstruction method has been applied to 16.5 GB LiDAR point cloud data of the suburb of Eltham in Melbourne and has successfully automatically generated thousands of building footprints in DXF format.

The LiDAR-based building extraction technique has been further developed and now works on complex buildings, such as multi-storey buildings and structures that have transparent materials.

Five papers were presented at the ISPRS Conference, Serving Society with Geoinformation. Two papers have been published in DICTA 2013 and another through the International Conference on Image and Vision Computing in New Zealand. A paper was also published in the ISPRS Journal of Photogrammetry and Remote Sensing.

Project Overview

Chunsun Zhang, Project Leader

...automated feature extraction from multi-sensor, multi-resolution and multi-temporal optical imagery and ranging data based on a new integrated data concept, the 3D image

Project Participants

Research & Education - Ergon Energy - University of Melbourne

43pl - AAM - Fugro Spatial - Geoimage - SKM - RPS

Government - DERM Qld - DSE Vic - Geoscience AustraliaLandgate WA - LPI NSW