Machine-vision software improves
By Ann R. Thryft, Contributing Technical Editor -- Test & Measurement World, 4/1/2009 2:00:00 AM
 Heiko Eisele President MVTec LLC |
Purchasing third-party machine-vision software can make a lot of sense for OEMs who don’t want to maintain their own in-house code library. Heiko Eisele, president of MVTec LLC, the US subsidiary of Germany-based MVTec, commented on recent changes in standard software packages and talked about how vendors are improving their offerings.
Q: What’s behind the rise of third-party, hardware-independent machine-vision software?
A: As machine-vision hardware has become a commodity, innovation now occurs mostly in software, and there are more vendors offering third-party standard packages. Standard products have been used in tens of thousands of similar installations and are extensively tested and field-proven, so they’re more reliable than a package you create in-house. There are still far more hardware than software manufacturers, but competition among software vendors has increased in terms of the tools, algorithms, and hardware SDK [software development kit] support they provide. For example, although MVTec has offered support for parallel processing since 2000, multiprocessor and multicore support is now also a feature of some other vendors’ software packages.
Q: What other improvements are appearing in machine-vision software?
A: Support for 3-D vision is improving. In 3-D calibration, for example, you want to be able to precisely measure the image, and that requires an accurate camera model. In our latest software release, Halcon 9.0, we extended the existing model to include higher-order mathematical terms. This gives a more accurate description of the imaging process, so the camera produces more accurate data. We also provided new operators for sheet-of-light measurement and added multigrid stereo capability.
Other changes include increasing the user friendliness of the development environment. For example, ours provides tools that let you develop algorithms much faster, by automatically generating code based on the user’s interaction with the software. Users have access to the entire library so they can write their own code, but certain tasks can be made more intuitive with a graphical user interface. We are providing this combination for more and more tools in our library, so users can quickly fix problems in the applications they develop.
Q: How else are vendors improving software?
A: In semiconductor and solar-cell inspection, as well as flat-panel displays, you must look at very fine detail in a large area. So, the use of line-scan cameras is increasing because of their higher resolution and their larger field of view. That means bigger images, which is the main reason we have removed the previous image size restriction of 32k x 32k. Larger images also require more memory, driving the trend to packages like Halcon that support 64-bit operating systems. As long as there’s enough memory in the host computer, you can process images of any size.
Another trend is the increased use of infrared imaging in inspection as sensor quality improves, bringing less noise and more accurate response. Infrared cameras used to be big and expensive, because the sensor had to be cooled down considerably to reduce image noise. Now, many infrared cameras operate at room temperature, making them cheaper and easier to use in a manufacturing environment, such as for managing heat dissipation.
No related content found.
- 0 rated items found.
Datasheets.com Electronic Parts & Inventory Search
185 million searchable parts
- Part Number
- Description
- Inventory
- Products
- Manufacturers
View All Webcasts
