DSL 3-D Imaging
Principle of DSL 3-D Imaging Operation
The theory of DSL 3-D (patent pending) measurement is based on quadric surfaces defined by a projected rotating grid pattern. The X, Y, Z coordinates of a point on a measured surface are calculated from the intersection of a pixel ray from the camera and a quadric surface. This method
- is scalable, from precision measurements of small parts to surface characterization of panels measuring 5 or 10 feet on a side
- can be used on objects with a wide range of surface brightness or reflection
- does not rely on assumptions inherent in, or implied by, parallel projection
- does not require an accurate measurement of the orientations and locations of system components at calibration time
- is generally applicable to any projection system in which projected points on a rotating grid sweep out quadric surfaces.
Using DSL 3-D System for Small Parts
The DSL 3-D system has been used to measure small parts (approximately 2 in. × 2 in.) with an accuracy of 0.004 in. The root mean square (RMS) error of a point cloud (76,000 points) on a gage block surface has been measured at 0.0012 in. Dense point clouds can be used to construct a surface mesh for dimensional inspection, reverse engineering, or surface rendering. The images below show a photograph of a small machined part and a wire-frame model of the part constructed with data recorded by the DSL 3-D imaging system. The data can be used for dimensional inspection in 3 dimensions. The wire-frame data can also be converted to a CAD file format (such as DXF) for use in reverse engineering applications.
Wire frame representation
The DSL 3D concept is being adapted to a microscope probe station for 3-D measurement of MEMS devices.
Using DSL 3-D System for Larger Area Measurements
Using shorter focal length lenses on the projector and camera allows DSL 3-D imaging to be used for measuring larger areas. Several aircraft control surfaces were measured with the camera and projector calibrated for a 4-ft × 8-ft field of view. Visual inspection (photo below, left) of a damaged flap shows a dent, but does not provide any quantitative measure of the damage. DSL 3-D imaging provides an accurate measurement for damage assessment. Below, right, is a wire mesh representation of the aircraft flap. The maximum depth of this dent calculated from the 3-D image was 0.118 in., which is in excellent agreement with the 0.150 in. measured with a dial indicator.
Flap with dent in surface
Wire frame representation of dented flap
DSL System Parameters
The design of a DSL measurement system involves selecting desired performance parameters of field of view, measurement resolution, and measurement time.
- The field of view can be scaled to fit specific applications by selecting appropriate lenses and specifying the standoff distance for the camera and the grid projector. A large field of view (for large objects) will require larger standoff distances, short focal length lenses, and a higher power light source. The standoff distance will typically be greater than the linear dimension of the object to be measured.
- The resolution of points measured in the field of view is determined by the resolution of the camera imaging array; typically, from several hundred to 1,000 pixels square for most machine vision cameras capable of high frame rates.
- The resolution that can be obtained in Z (elevation) depends on both the field of view and the number of images recorded for processing. The highest resolution is obtained for small fields of view and a large number of recorded images.
- Measurement time includes time for acquiring a sequence of images of the object and time for computing the elevation at selected points on the surface. The image acquisition time depends on the number of image frames recorded and the frame rate of the camera.
For a given field of view, higher resolution can be obtained by acquiring data for longer periods of time. The plot below shows theoretical measurement resolution vs. image frames recorded for measurement areas (field of view) ranging from 0.25 in. to 8 ft. For example, measuring a resolution of 0.003 in. over a 1-ft. × 1-ft. area would require approximately 3,000 recorded frames for an acquisition time of about 50 sec. at 60 frames per second (fps) or about 6 sec. using a 480-fps camera.
Graph showing theoretical resolution by FOV and image frames recorded
The process of generating a 3-D image includes two distinct steps:
- With a part in position in the field of view, image frames are transferred from the camera as the grating rotates. Initial image processing is conducted in real-time as images are acquired from the camera. Image acquisition continues while the grating rotates through at least one revolution to obtain a full cycle of the projected pattern. After all images are acquired and initial processing results are stored in memory, the part can be moved from the measurement station while the elevations at selected grid locations on the surface are calculated.
- With all images acquired, the second step begins—coordinates of selected points are computed from the intermediate processing results. Using a 1-GHz Pentium computer as the hardware platform, the coordinates can be computed at a rate of approximately 20,000 points per second. Reverse engineering applications might require evaluation of a dense point cloud (300,000 points), which would require a computation time of 15 seconds using the 1-GHz platform. Lower density sampling may be appropriate for many applications, and these would have correspondingly shorter evaluation time periods.
digital imaging software • 3-D imaging • reverse engineering • rapid prototyping • 3-D measurement • surface roughness measurement • surface profile measurement • automated optical inspection • non-contact surface profile • thickness measurement • equipment surface angle measurement • surface measurement systems • non-contact surface finish measurement • non-contact automated dimensional inspection • three dimensional images • 3-D images • dynamic structured light • noncontact surface measurement • optical measurement • lower cost • higher accuracy • scalable microscopic • large area custom • system development • long stand-off