A DIGITAL PHOTOGRAMMETRIC SYSTEM FOR THREE-DIMENSIONAL
DEFORMATION MEASUREMENT
Weiyang Zhou, Robert Brock, James Thorpe and Paul Hopkins
SUNY College of Environmental Science and Forestry
Syracuse, NY 13210
ABSTRACT:
A self-calibrating digital photogrammetric system is employed to determine three dimensions in object space. Digital
imagery is collected by two Panasonic WV-CD20 CCD cameras at a distance of about 26 inches from the object specimen.
The image coordinates of control placed in the object space are measured from a monitor. The image coordinates of random
dots placed on the specimen are determined using a feature-based image matching procedure. A photogrammetric bundle
adjustment method constrained by weights on appropriate variables provides the solution for all parameters including the
object coordinates of the random dots on the specimen. Results to date indicate that average standard deviations in inches for
the random dots are 0.006, 0.006, and 0.027 respectively, for X, Y, and Z.
KEY WORDS: digital photogrammetry, close-range, self-calibrating, stereo-pair image matching, 3-D reconstruction,
bundle adjustment.
1. INTRODUCTION
In solid mechanics the surface of a specimen will deform
when force is applied to it. The present analysis of this
deformation utilizes a linear image strain analysis system,
LISA, which is able to detect and measure the displacement
in two orthogonal directions (X and Y) in the plane of the
surface of the specimen. Usually the measured directions
are across and along the direction of applied force.
The purpose of an ongoing project is to develop a
non-contact digital image acquisition and analysis system
to quantify the deformation in three dimensions. This paper
reports on the progress of this project to date.
2. BACKGROUND
The principles of photogrammetry have long been used in
non-topographical measurement [4] [7]. With the advent
of digital sensors digital images are being used as
information sources instead of photos. In comparison with
traditional hard copy photos digital images have the
advantage of being highly computer-compatible. With the
ever-increasing power and availability of computers and
data sources, and the increasing need for real time
processing, digital image processing and analysis has
become an inseparable part of analytical photogrammetry.
By using digital cameras some of the errors associated with
the traditional methods can be avoided, such as those
introduced by the deformation of the film negative during
the procedure of developing and drying.
The use of non-metric digital cameras requires that a
self-calibration procedure be used. Generally CCD
(charge coupled device) cameras have no fiducial marks
and the location of the principal point must be calculated.
The CCD camera may use different lenses for different
applications and thus require calibration for each use.
Likewise the principal distance of the CCD camera used for
close-range work is constantly changing and its
determination must be made for each image processed.
Often in the use of self-calibrating bundle adjustments the
camera station parameters are of no real interest [7].
Generally one is concerned with the object point
coordinates and their error estimates.
Most photogrammetric techniques are based on the
geometric relationships between the object's position in the
object space and its position on the surface of the sensitive
recording part (i.e., the negative film with the traditional
camera and the CCD chip with the CCD camera) of the
sensor. The position of an object point, after projection
onto the surface of the sensor of the CCD camera, is
unknown and cannot be measured directly as with the
traditional negatives. The position of the image in the
sensor plane must be determined in terms of pixel
coordinates. The actual space represented by one pixel is
critical to the success of the data reduction. The actual
ability to measure coordinates in CCD cameras is often 4
0.015 mm. whereas this typically is + 0.005 mm. in terms of
film based cameras. This difference between the camera
systems in coordinate measurement capability will
decrease as CCD technology advances.
3. THE EXPERIMENT
3.1 Objective
The objective of this study is to develop a digital self-
calibrating analytical photogrammetric procedure to
determine the three-dimensional spatial coordinates of
random points on a surface whose image positions are
automatically determined through image matching
techniques. Figure 1 shows a schematic diagram of the
experimental setup.
32 Hardware
The layout of the computer image analysis system is as
shown in Fig. 2. It consists of the components listed below.
1. The host computer. A SUN SPARC system 330 with
16Mbyte memory, SunOS 4.1 operating system and a 1/4 in
tape drive.
