Analysis of and Feature Extraction from Dense 3D Point Clouds Generated from Imagery, Radar and LiDAR.
Significant and ongoing increases in the volume, resolution and accuracy of remotely sensed data being acquired for the geospatial-intelligence community has led to the need for enhanced levels of automation in the generation of spatial information products. Dense point clouds facilitate the detection, extraction and modelling of features of interest, be they man-made or natural, and perhaps more importantly they facilitate the detection and temporal monitoring of changes within a scene. The 3D reconstruction and modelling workflow, from pre-processing through the processing pipeline to interactive display of reconstruction, has traditionally involved a large component of manual, labour-intensive processing. This project aimed to both investigate innovative new approaches to 3D point cloud processing and develop software tools to demonstrate the new capabilities.
The project partner was Australian Geospatial-Intelligence Organisation (AGO).
This project was an extension of a previous CRCSI project entitled Enhancement of Close-Range Photogrammetry Technology for Defence and National Security Applications. After the initial 12 months, this project was extended for a further 12 months.
A 3D point cloud in this context represents a surface model with irregularly distributed points, which can be both very dense and very accurate, but also potentially contaminated with noise. There was an ongoing need for innovative new techniques and computational processes to enhance automated generation of 3D point clouds, especially from imagery. Further developments in automatic detection and extraction from point clouds of object features in vectorised form, to enable geometric characterisation, modelling and analysis, was also warranted.
The aims of the first 12 months of the project were:
- To investigate both current and potentially new analysis and processing tools to enhance 3D point cloud generation, object feature extraction and modelling from dense point clouds.
- To develop a suite of software tools to perform point cloud generation and processing in order to demonstrate utility and functionality. The development focussed on functions of high priority to the AGO, especially in the area of photogrammetric spatial data generation from non-traditional image sources.
The 12 month extension project continued these aims, with a focus on the integration of data from imaging and ranging sensors on space-borne, airborne and terrestrial platforms through new data fusion tools as a key prerequisite for comprehensive implementation of Activity-Based Intelligence (ABI). ABI is an analysis methodology which rapidly integrates data from multiple sensors and intelligence sources relating to the interactions of people, events and activities, in order to discover relevant patterns, determine and identify change, and characterize those patterns to drive collection and create a decision advantage.
This project enabled advances in automation for the AGO, improving efficiency and reducing the need for manual processing. It also enhanced analysis techniques for image-based point cloud generation and classification, temporal analysis tools for change detection, and 3D reconstruction and information extraction from full-motion video. The software tools delivered demonstrated the functionality of the recommended algorithms and computational schemes to the AGO and they are now being utilised in production operations. A future project would allow further development of the tools and associated software package for 3D point cloud generation and processing.
To learn more about this project, contact us at email@example.com