“Structured light” can refer to a number of 3D scanner designs that project various light patterns onto an object, and use either single or stereo cameras to decode the reflected images to generate a 3D point cloud.
Projections like random dots or codewords (repeated mini-patterns) generate lower XY resolution, but can deliver higher frame rates than fringe projection since only a few projections are needed to produce 3D. Handheld scanners use this approach and average many frames to improve XY resolution using a fusion algorithm.
Fringe Projection for High-Resolution Scanning
To achieve the highest resolution scanning, however, fringe projection is the most common. In this technique, sinusoidal patterns are moved across the field of view to achieve sub-pixel performance. Fringe projection generally requires 10 or more patterns to produce 3D and is therefore slower than lower resolution projection techniques.
Aside from projection, the main differences in sensor design are in the use of a single or stereo cameras.
Single Camera Technologies
Over the past year, there have been several products entering the market offering single camera based 3D. Some examples are Chiaro (http://chiarotech.com, recently acquired by Cognex), Zivid Labs (http://www.zividlabs.com), and Zeiss’ AIMax Cloud (http://www.zeiss.com).
The Zivid 3D camera for industrial automation.
Stereo Camera Advantages
The differences in a single camera approach over stereo are subtle but important.
Stereo cameras see more data and therefore produce less occlusions. With a single camera, any occlusion caused by surface geometry that blocks projected light or the camera’s view will result in no data. Stereo cameras, on the other hand, effectively use three views of the part to capture more data –– triangulation of the left camera, triangulation of the right camera, and stereo with two cameras working together.
Stereo camera design minimizes occlusions.
Occlusions can be a significant roadblock to achieving reliable and accurate feature measurement. If you can’t see a portion of the feature, then your data may be incorrect. Stereo designs offer a greater chance to make accurate measurements from a single acquisition sequence.
The calibration approach for single camera and stereo cameras is also very different. Single camera designs must be calibrated to establish a triangulation relationship between the camera and projector. If the projected pattern moves due to temperature changes as the light source heats up, then the 3D data produced will be incorrect.
In contrast, stereo cameras are calibrated between the camera pair and not with respect to the projector, so movement in patterns caused by temperature changes do not result in erroneous 3D data.
In conclusion, stereo technology produces higher data quality and repeatability as a result of less occlusion and better temperature stability. This is why Gocator uses a stereo design in addition to fringe projection to deliver the highest resolution in XYZ and hardware acceleration to achieve 4 Hz full-field snapshot performance.