An improved method of dot-projection photogrammetry and an extension of the method to encompass dot-projection videogrammetry overcome some deficiencies of dot-projection photogrammetry as previously practiced. The improved method makes it possible to perform dot-projection photogrammetry or videogrammetry on targets that have previously not been amenable to dot-projection photogrammetry because they do not scatter enough light. Such targets include ones that are transparent, specularly reflective, or dark.

In standard dot-projection photogrammetry, multiple beams of white light are projected onto the surface of an object of interest (denoted the target) to form a known pattern of bright dots. The illuminated surface is imaged in one or more cameras oriented at a nonzero angle or angles with respect to a central axis of the illuminating beams.

The locations of the dots in the image(s) contain stereoscopic information on the locations of the dots, and, hence, on the location, shape, and orientation of the illuminated surface of the target. The images are digitized and processed to extract this information. Hardware and software to implement standard dot-projection photogrammetry are commercially available.

A Laser Beam Is Split into multiple beams to form dots of illumination on a target. The dots are observed by use of stereoscopic cameras.

Success in dot-projection photogrammetry depends on achieving sufficient signal-to-noise ratios: that is, it depends on scattering of enough light by the target so that the dots as imaged in the camera(s) stand out clearly against the ambient-illumination component of the image of the target. In one technique used previously to increase the signal-to-noise ratio, the target is illuminated by intense, pulsed laser light and the light entering the camera(s) is band-pass filtered at the laser wavelength. Unfortunately, speckle caused by the coherence of the laser light engenders apparent movement in the projected dots, thereby giving rise to errors in the measurement of the centroids of the dots and corresponding errors in the computed shape and location of the surface of the target.

The improved method is denoted laser-induced-fluorescence photogrammetry. In preparation for observation by this method, a target is made fluorescent by whichever of several techniques is most appropriate. Examples include the following:

  • A target sheet or membrane (e.g., a membrane subject to deflections that one seeks to measure) made of a transparent or translucent polymer could be doped with a fluorescent dye during its manufacture. If, in addition, the target were coated on one side with metal to render it specularly reflective,then it could be positioned for photogrammetric observation of its uncoated side.
  • A thick transparent, translucent, or dark opaque object (e.g., a wind-tunnel model, the position, orientation, and deflection of which one seeks to measure) could be doped with a fluorescent dye during its manufacture.
  • A specularly reflective or dark object that has already been manufactured could be coated with a paint or lacquer containing a dissolved fluorescent dye.
  • The surface of a part of a human body to be observed photogrammetrically could be coated painted or sprayed with a harmless fluorescent dye in a nontoxic solvent.
  • A fluorescent dye could be dissolved in a body of liquid, which could be, for example, water in a tank for experiments on the sizes, shapes, and motions of waves.

In this method, a laser beam is pulsed and is split into multiple beams by use of a commercially available diffractive optical element (see figure). The laser beams form dots of light on the target, exciting fluorescence in the dye. Unlike the laser light, the fluorescence from the dots is incoherent and therefore, there is no laser speckle in the fluorescence. Like scattered light, the fluorescent light is emitted in all directions.

The cameras are time-gated to acquire images only during the interval of maximum fluorescence following the laser pulse, to suppress the contribution of ambient light between pulses. To suppress ambient light further, the light entering the cameras is band-pass filtered at the fluorescence wavelength, which is longer than the laser wavelength. This filtering also blocks directly reflected laser light, which, if allowed to enter the cameras, could give rise to spurious dot images.

The extension of the method to videogrammetry is straightforward: The laser and cameras can be operated in a sequence of pulses to acquire a sequence of projected-dot images. The images, or the sets of location and shape data extracted from the images, can be used to synthesize a motion picture showing how the target surface moves and/or deflects over time.

This work was done by Paul Danehy, Tom Jones, John Connell, Keith Belvin, and Kent Watson of Langley Research Center. LAR-16426


Photonics Tech Briefs Magazine

This article first appeared in the July, 2004 issue of Photonics Tech Briefs Magazine.

Read more articles from the archives here.