School of Surveying and Spatial Information Systems

 

The University of New South Wales

 


 

Application of PhotoModeler Pro 5 Software

 

 

by Lee Schmalfeldt

 

Supervised by B. Donnelly

 

 

Edited by J. M. Rüeger

 

October 2003


 

Introduction

 

The first photogrammetric methods were developed in 1851 by a Frenchman named Aime Laussedat. Laussedat’s method involved using terrestrial photographs that, when taken from different camera stations, formed intersecting rays on which measurements could be made. Since Laussedat first formulated the concept of photogrammetry, it has moved forward due to advances in technology. The current phase of photogrammetry is digital photogrammetry. Digital photogrammetry has come about due to the increasing power of computers as well as the invention of the digital camera. A number of digital photogrammetry programs are currently available on the market. One such program is PhotoModeler Pro 5. PhotoModeler is a MS-Windows based digital close-range photogrammetry program that allows 3-D models to be constructed from digital images. PhotoModeler is primarily designed for users with limited experience in the field of photogrammetry. The application of PhotoModeler has been tested by undertaking two projects, namely producing a 3-D model of the facade of a building and producing a 3-D model of the top of a racing paddleboard. The first project aimed at testing the standard processing tools in PhotoModeler whereas the second project aimed at testing the automated processing tools in PhotoModeler. A Sony DSC-P71 consumer grade digital camera was used to record the images for both projects. This camera was calibrated using the camera calibration facility in PhotoModeler and has a maximum resolution of 3.32 megapixels.

 

 

Building Facade Project

 

Figure 1 shows the building that was used for the project. Images of the building were recorded so that all features of interest appeared in a least three images. Before starting the processing in PhotoModeler, the basic tutorial videos, that came with PhotoModeler, were viewed. These videos covered all of the elementary processing tools used to produce a 3-D model in PhotoModeler and were extremely helpful. Once the familiarisation stage was complete, the images were processed in PhotoModeler. The processing stage involved marking and referencing the features of interest on all images. No formal targets were placed on the building. Once all  features had been marked and referenced, surfaces were added using the surface tools in PhotoModeler. Figure 2 shows the final 3-D model with all surfaces added.

 

 

 

Figure 1 Building used for the project.

 

 

 

Figure 2 Resulting 3-D model.

 

The model was scaled by specifying that two points were a given distance apart. The accuracy of the model was then tested by comparing 30 distances on the scale model with the same 30 distances measured on the building using a 10 m tape. The 30 distances covered the span of the building and included distances in the horizontal (x), vertical (z) and depth (y) directions. From the comparisons a RMS value for the distances on the scale model was determined to be ±7 mm, or ±1.6 pixels at the image scale.

 

 

Paddleboard Project

 

This project aimed at testing the automated processing tools in PhotoModeler. The Paddleboard is an irregularly shaped object that is difficult and time consuming to model using conventional methods, such as theodolite intersections. The images were recorded in a dance studio at night to eliminate the amount of ambient light. Targets were positioned around the paddleboard to be used as relative control points to orient the images. Circular retro-reflective targets were placed on the paddleboard to represent the general shape of the top of the paddleboard. Figure 3 shows the paddleboard used for the project.

 

 

 

Figure 3 Paddleboard used for the project.

 

Once the images were oriented, using the targets positioned around the paddleboard, the circular retro-reflective targets on the paddleboard were marked using the automatic target-marking tool. All that was required was to specify the extent of the area to be marked, the largest anticipated target diameter in pixels, how circular the targets had to be in order to be marked and which target marking method was to be used. Once the marking was completed, the targets were referenced using the automatic referencing tool. This tool identifies points that are the same on different images, from the initial orientation, and references them as being the same. From the referenced points it was possible to produce a triangulated irregular network (TIN) of the paddleboard. A TIN produces a 3-D model by forming triangles between all referenced points. Each of these triangles then becomes a sub-surface of the entire TIN surface. Figure 4 shows the TIN model of the paddleboard produced in PhotoModeler.

 

 

 

Figure 4 TIN model produced in PhotoModeler.

 

 

Conclusion

 

The two projects proved that PhotoModeler is suitable for users with limited experience in close-range photogrammetry. The accuracy of PhotoModeler as a measuring tool was significantly better than expected. The anticipated accuracy of the scale model of the facade was in the order of a few centimetres. An RMS value of ±7 mm indicates that PhotoModeler is capable of accurate measurements, even when no formal targets are used. The automated tools in PhotoModeler proved to be easy to use and very efficient. The automated tools allowed the paddleboard project to be completed far more quickly and accurately than if manual marking and referencing was undertaken.

 

One problem with PhotoModeler is the amount of random access memory (RAM) required to run the software without it crashing. The computer used to run PhotoModeler had 256KB RAM, but still crashed on a regular basis. Another problem is that much of the mathematical processes in PhotoModeler are hidden from the user. This is suitable for users with limited close-range photogrammetry experience, but not for more experienced users who wish to control the processing.

 

It is recommended that:

*        the accuracy of PhotoModeler as a measuring tool be further tested against a method capable of obtaining a higher order of accuracy,

*        the calibration facility in PhotoModeler be tested using a metric camera,

*        the mathematical model used in PhotoModeler be further investigated, and

*        PhotoModeler be used as a teaching tool to explain the elementary concepts of photogrammetry.

 

Further Information

 

For more information, please contact:

Brian Donnelly (Supervisor)

Email:  B.Donnelly@unsw.edu.au

 

Mail:

School of Surveying and Spatial Information Systems

University of New South Wales

UNSW SYDNEY NSW 2052

Australia

 

Phone: +61-2-9385-4202

Fax: +61-2-9313-7493

WWW: http://www.gmat.unsw.edu.au