Journal of Surveying Engineering

An attempt was made to investigate the state-of-the-art use of a digital and near real-time system to monitor monuments and historical sites by integrating semiautomated PC-based computer vision and stereo-photogrammetric techniques. The effects of the number and type of control, particularly three-dimensional point location and distance, on the reliability and effectiveness of camera calibration procedures using planar objects as a control facility were studied. The trends of interior and exterior calibration parameters were investigated using 25 combinations of different numbers of measured distances and 3-D control points, and with a fixed focal length for the calibrated charge-coupled device (CCD) cameras. Results showed that increasing the number of distances and 3-D control points above the minimum required control numbers had a slight effect on the values of interior geometry parameters of CCD cameras as well as on the planar object coefficients. The consistency of all test results, except those with control greater than 10 distances and 20 control points, was the major reason for choosing the minimum amount of control for calibration. Quantification of the standard error of image coordinate residuals, distances residuals, standard error of unit weight, and number of iterations for convergence of algorithms supported this result. A historical statue was mapped using single camera operation and a multiple model mapping approach to reconstruct a digital 3-D surface model and thus compute the statue's geometrical characteristics. Three stereo models of the scene were captured from different perspectives. A movable cubic control facility located in the field of view of the mapped scene was used to scale the stereo models. A digital model for the statue was constructed, along with cross sections taken at selected height intervals. The mapping and camera calibration procedures were simple and practical.

The use of a shore-based pseudolite to augment DGPS for on-the-fly (OTF) carrier phase ambiguity resolution in the marine environment is investigated. OTF ambiguity resolution results in cm-level tridimensional accuracies. This high level of accuracy is required for applications such as depth determination in shallow navigation channels, dredging, and harbor maneuvers. Pseudolite augmentation results in improved satellite geometry and more effective and reliable ambiguity resolution. The use of a pseudolite can be crucial to resolving the ambiguities in areas where signal masking occurs due to topographic or man-made features. A shipborne test was conducted on Lake Okanagan, British Columbia, in November 1996, to validate these claims and assess pseudolite performance under field test conditions. A Stanford Telecom L1 pseudolite (Model 7201) was used, together with two NovAtel OEM-2 (RT-20) L1 12-channel Narrow Correlator receivers, which can receive the STel pseudolite signals without any modifications. The distance from ship to shore, where the reference receiver and pseudolite were located, reached 3,250 m. Between five and eight satellites were available during two tests. When using the full GPS constellation and with a masking angle of 10°, it was found that pseudolite augmentation reduced the average time to ambiguity resolution by 35 to 40, while improving the percentage of time when the ambiguities were solved correctly in the initial attempt from 95 to practically 100. Weak geometry was simulated by removing a satellite from the data processing. Under such a condition, pseudolite augmentation reduced the average time to ambiguity resolution by more than 50 and resulted in a significant improvement in ambiguity resolution reliability.

The California Department of Transportation (CALTRANS) conducted a field test of the Trimble 4000 SSE Real-Time Kinematic (RTK) GPS survey system in February and June of 1996. The object of the test was to independently analyze the repeatability of the RTK survey system. During three days of observations, 156 station occupations were performed on a network of 10 stations. The results of the RTK testing were then compared against results from a static GPS survey of the network. Ellipsoid heights from the static survey were combined with the GEOID93 height model and compared against geodetic leveling as a further quality check. All RTK observations were limited to no more than 500 m in length. In these comparisons, RTK results demonstrated RMS errors of ±12.2 mm and ±22.6 mm for their respective horizontal and vertical components. Evidence is presented that suggests that vertical RTK accuracy may be affected by time-dependent systematic error.

As a continuation of the writers' previous work on establishment of Tienstra's formula via finite-element ideas for an internal observation point, the formula is proven for the case where the observation point lies outside the control triangle. Because of the loss of “rotational symmetry” in the present case, this proof differs from the previous one as more formulas have to be derived, and “reflection symmetry” of the new configuration becomes an important aid throughout the derivation. Furthermore, the lengthy computer algebra needed to complete the previous proof has a much simpler counterpart here, requiring only one line of computer simplification.