Ground-based hyperspectral imaging for geological outcrop analysis
The limitation of photorealistic 3D models
The ability to represent geological outcrops with high geometric accuracy has made terrestrial lidar scanning a powerful tool for geological outcrop analysis. Many modern lidar-based geology projects integrate data from high resolution digital cameras to create photorealistic 3D models, which are used to map and interpret geological features. However, most of the interpretation is carried out manually and the mapping is limited to features which can be detected from the lidar geometry or from the visible wavelength images.
Geological interpretation of a photorealistic 3D model.
What is hyperspectral imaging?
Hyperspectral imaging has been used for geological applications for many years, from airborne and spaceborne platforms, and is based on the physical interaction of electromagnetic radiation (reflection properties of sunlight) with surface material. The reflection and absorption of certain wavelengths is directly controlled by the chemical composition and the crystal structure of the object. Many minerals show specific absorption features in the visible and infrared wavelengths, which can be used as a diagnostic tool to identify and map these mineral and chemical variations. With hyperspectral scanners, the reflected light is sampled with a high number (several hundred) of very narrow (~5nm) spectral bands, which results in a near-continuous spectral curve for each image pixel.
The hyperspectral data cube: the high number of very narrow spectral bands results in a continuous spectral curve for each image pixel. The front of the cube shows a false colour image using the infrared spectral bands 1721nm, 2306nm and 1565nm in RGB.
Close range hyperspectral imaging - a new area in remote sensing
To scan near-vertical cliff sections of geological outcrops, a close range instrument is necessary. The aim of this project is to integrate terrestrial laser scanning with ground-based hyperspectral imaging to complement the 3D lidar models with quantitative information about the chemical and mineral variations in the outcrop, and to enable a more automated outcrop evaluation. The extreme off-nadir (nearly horizontal) scanning view and the resulting scan geometry need to be taken into account during processing.
The project focuses on:Workflow development to implement spectral images in virtual outcrop analysis.
Photogrammetric integration of hyperspectral images with the 3D lidar models.
Unconformity corrections of ground-based hyperspectral images.
Extracting and mapping of geological features from close range hyperspectral images.
Testing the method in carbonates and siliciclastic settings.
Close range hyperspectral imager.
False colour image with spectral bands 1721nm, 2306nm and 1565nm visualised in RGB.
Mixture Tuned Matched Filtering result of a processed spectral image. The images show the relative abundance of limestone, dolomite and vegetation for each pixel.
Photorealistic 3D model integrated with a hyperspectral classification image.