This thesis develops the potential of consumer-grade digital cameras for accurate
spatial measurement. These cameras are generally considered unstable but
their uncertain geometry can be partially resolved by calibration. The validity
of calibration data over time should be carefully assessed before subsequent
photogrammetric measurement. The use of such digital cameras for photogrammetric
measurement is increasingly accepted in many industrial fields but also in
a diverse range of fields including medical and forensic science and architectural
work. However, the stability of these cameras is less frequently reported in the
literature, which can be attributed to the absence of standards for quantitative
analyses of camera stability.
The approach used to assess camera stability in this study is based on comparing
the accuracy in the reconstructed object space, achieved using sets of interior
orientation parameters of a sensor, derived in different calibration sessions. This
technique was successfully applied to assess the temporal stability and manufacturing
consistency of seven identical Nikon Coolpix 5400 digital cameras. These
cameras demonstrated remarkable potential to maintain their internal geometry
over a 1-year period. This study also identified residual systematic error surfaces,
discernable in digital elevation models (DEMS) derived from image pairs. These
’domes’ are caused by slightly inaccurately estimated lens distortion parameters.
A methodology that uses a mildly convergent image configuration removes
the systematic error sources. This result is significant for DEM generation using
low-cost digital cameras and a series of case studies demonstrated that this
methodology can reduce the need for an accurate lens model and effectively increase
the accuracy achievable with non-metric digital sensors.
A Doctoral Thesis Submitted for the Degree of Doctor of Philosophy.