UCL Discovery
UCL home » Library Services » Electronic resources » UCL Discovery

Procedures for the analysis and use of multiple view angle image data

Kay, Simon A W; (1990) Procedures for the analysis and use of multiple view angle image data. Doctoral thesis (Ph.D), UCL (University College London). Green open access

[thumbnail of Procedures_for_the_analysis_an.pdf] Text
Procedures_for_the_analysis_an.pdf

Download (12MB)

Abstract

It is recognised that the majority of vegetative cover types have anisotropic reflectance characteristics that are largely a function of their canopy geometry. Several studies have made attempts at formulating methods for the use of data remotely sensed from off-nadir directions. The best of these methods attempt to utilise the "extra" information implicitly contained in off-nadir image datasets. In this study, an attempt is made to extract information concerning agro-physical parameters of a number of vegetative cover types using imagery acquired by an airborne sensor, the Daedalus Airborne Thematic Mapper (ATM). It is also recognised in the literature that the nature of spatial variance in images is related to the size and distribution of the objects in the scene and the sampling characteristics of the sensor. In previous work this relationship has been explored by examining scenes using images of varying spatial resolutions, using a number of measurements of spatial variance. The underlying trend of these measurements is then used to interpret the nature of the objects in the scene. No previous work exists which attempts to utilise the change in variance of images acquired at different off-nadir view angles. In this study, the understanding of this relationship is developed by examining the change in variance of a number of vegetative cover types from multiple view angle image datasets. The geometry of the ATM sensor is derived to allow an understanding of the sampling characteristics of the instrument. Two important geometric factors are established: first, the area of the ground resolution element increases with view angle, which effectively reduces spatial resolution at off-nadir angles; and second, overlap between adjacent ground resolution elements increases with view angle, increasing the spatial auto-correlation between these samples. The effects of illumination, atmosphere and topography can all influence variance in an image. A parametric procedure for normalising multiple view angle (and therefore multitemporal) datasets for these factors is developed, based upon the production of reflectance images using a sky radiance model of the spectral and spatial distributions of irradiance, ground measurements of irradiance, and a digital terrain model of the study site. Finally, it is shown that image variance is likely to decrease at off-nadir view angles, the magnitude of this decrease being related to the sensor geometry and (more importantly) the geometry of the canopy. By a simple statistical analytical procedure it is possible to construct broad classes within which the nature of the canopy can be classified.

Type: Thesis (Doctoral)
Qualification: Ph.D
Title: Procedures for the analysis and use of multiple view angle image data
Open access status: An open access version is available from UCL Discovery
Language: English
Additional information: Thesis digitised by ProQuest.
Keywords: Earth sciences; Vegetative cover
URI: https://discovery.ucl.ac.uk/id/eprint/10123988
Downloads since deposit
0Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item