Close-range Modelling of Table
Objective
This case study evaluated ObliqueMapper's ability to collect accurate features of an object at close-range (within a few metres). A table consisting of several geometric surfaces (generally Z-shaped) was chosen as the subject; and was ultimately digitized as a 3D model.
Photo Acquisition
The table was placed in the middle of the room, and eight photographs were taken at multiple angles circling the table at a distance of about 2-3 metres. For each photograph, the camera was kept relatively level and aimed to keep the table in the center of the frame.
GCP Collection
Since the area of interest was relatively small and measurements on the order of millimetres was expected, GPS was not suitable to establish control for the photographs. Instead, a Leica Total Station (a laser ranger capable of millimetre accuracy) was used to collect the positions of five targets that were placed on the floor, walls, and top of the table.
In addition, to allow coincident points to be compared between the total station and ObliqueMapper, several smaller targets (65 in total) were placed on various surfaces of the table and then measured.
Feature Collection
ObliqueMapper was used to collect 3D locations of the same 65 targets collected by the total station. Several targets were collected multiple times from different total station setups and ObliqueMapper photographs to help assess the repeatibility of measurements.
Results
Comparison of the 65 target coordinates acquired using the total station with those determined by ObliqueMapper showed agreement within 2.5 mm (95% confidence). For targets that were measured more than once, the standard deviations of the total station measurements were on the order of 2 mm (95% confidence), with ObliqueMapper in the range of 3 mm (95% confidence).
Although not initially planned for this case study, a 3D model of the table was created from the corners, edges, etc. that could be seen throughout the eight photographs (taking about 5 hours). The quality of the model also helps to show ObliqueMapper's ability to capture detailed models of objects at close-range.