The evolution of vision systems
I believe we can safely say that vision systems will become more and more important in the future.
Security, traffic, food, health products and medical procedures as well as production quality control will all require sophisticated vision systems and cost will continue to be a factor.
What if we could have a vision system that is small, easy to implement and affordable for every user?
The answer is a smart camera, a fully integrated device that includes a sensor, signal processing, image processing and a complete application layer useful for any inspection.
The smart camera of the future must be capable of providing adaptable video filters such as noise reduction, dynamic range manipulation, color space conversions, cropping, space rotation, pan, tilt, and auto correction for bad images or poor illumination.
Once raw video has passed through pre-processing, image processing takes over with video analytics, motion detection, object tracking, on-the-fly locators, pattern matching, image recognition etc..
After all this signal and image processing, software will provide inspection of objects, measurement for quality controls, pass, recycle, reject solutions and a script language that allows a user the flexibility to write his own control sequence or even create some image processing functions.
In order to produce such a camera, we would need hardware that is small, and low power with many interfaces built in a single chip. Perhaps wireless hardware is the way to go. Many of the wireless components use multi-ARM processors with FPGA’s as real-time signal processors along with a rich support of hardware peripherals. Then, a powerful OS, Linux and Android are good examples, among others.
All this hardware and software must be integrated into one camera. In addition, the camera must be able to handle vibration, wide ranging temperatures and hot plugging. Users are ruthless and environments are harsh, reliability and robust designs are critical.
The advantages of this dream smart-camera-vision system are enormous.
- Requirement for less space and fewer cables. Interface is direct to host computer.
- Low number of components so the MTBF is inherently better.
- Lower cost, easier troubleshooting and lower maintenance.
- Easy integration due to on-camera intelligence and flexibility.
- Remote configuration and support.
Today’s generation of smart cameras, including Teledyne DALSA’s BOA family, demonstrate how much closer we are to this goal now than we were 5 years ago – can we get all the way there in the next 2 to 3?
