Robert McEwan
Ìý
B.Sc. (Honours) Thesis
(PDF - 3.6 Mb)
An area of leucomonzogranite, north of Herring Cove, Nova Scotia provides a well exposed section of nearly continuous outcrop used as a test of a remotely sensed image analysis. The objective is to examine the effectiveness of a charge coupled device (CCD) in producing whole rock spectra based on average pixel values, and in the identification of absorption features in hand samples. Three test sites were imaged using a near-infrared (NIR) converted Nikon D40 SLR camera with filters in the blue, green, red, and NIR range. Samples from each test site were imaged at a fine (mm) scale using a bellows, with the same filters as the field test sites. A NIR removal method was determined to correct color filtered images with additional transmission in the NIR region. Image analysis revealed that spectral variation in the NIR region is controlled by biotite and quartz content, which occur as millimeter sized clusters in outcrop samples. An NDVI calculation was used to separate lichen and surface depression pixels from outcrop pixels to compile a spectra for each outcrop site.
Error in the results are reported as wavelength specific over-compensation and under-compensation from the NIR removal process, and non-linear detection of NIR light on the sensor versus exposure time. Overexposure and camera drift are field specific sources of error.
Keywords: Halifax pluton; remote sensing; charge coupled device; image processing; multispectral analysis; absorption bands; imaging spectroscopy
Pages: 54
Supervisor: Charles Walls