A simpler camera with embedded intelligence sounds like a good idea for machine vision. In fact, as editor Andy Wilson writes in his My View column in this issue, many vendors of machine-vision cameras are already heading in this direction, adding FPGAs, DSPs, and GPUs to their products so that their customers can build ever more sophisticated systems.

Researchers at Carnegie Mellon University’s National Robotics Engineering Center have developed a plant-sorting machine that uses machine vision and image processing. The machine inspects and grades harvested strawberry plants, and then mechanically sorts by quality—tasks that until now could only be done manually.

In surveillance, a user-friendly and robust automated object detection and tracking system that can handle complex situations is not yet available. Recently, Oliver Sidla and his colleagues at SLR Engineering have taken a different approach from those underlying most current surveillance systems. Instead of generating a sophisticated background frame and using pixel differences between images, their system discovers pedestrians solely based on their appearance by learning what people look like.

Jason Fleischer and his colleagues at Princeton University’s Nonlinear Photonics Group are developing an imaging technique that relies on a nonlinear crystal rather than a standard lens. The research, funded by the Air Force Research Laboratory, may prompt advances in the areas of data encryption and wide-area, high-resolution photography.

Metrology consulting firm Pine Falls Technical Services has shown the Royal Canadian Mounted Police (RCMP) and the Winnipeg Police Department how a laser-based imaging device could piece together the remains of a vehicle that had been destroyed by a bomb.

A number of different methods exist to generate 3-D image data. These include structured laser light-based systems, time-of-flight (TOF)-based methods, and projected structured light systems. Structured light projection using laser-based systems is often used to create high-resolution 3-D images; however, scanning the object requires that the object or camera system move across the object’s field of view. Alternate methods, such as TOF systems, can generate these 3-D models using a single camera with an embedded light source, although 3-D image resolution is far less than that of laser scanning-based systems.

In many machine-vision systems for dimensional metrology, laser lines are used to provide structured illumination—allowing height to be estimated through triangulation using the known relative orientation between camera and laser line source. By projecting laser lines onto a moving object, or mounting camera and laser line source onto a moving robot arm, point clouds of data can be obtained and used to reconstruct a 3-D model of the object.

One of the most important functions of any machine-vision system is pattern matching. In vision-guided robotics, for example, pattern-matching algorithms are widely used in pick-and-place systems where camera-guided robots need to pick and then place randomly oriented parts. Indeed, so important is pattern matching that specific algorithms to perform the task have been the subject of multiple lawsuits from software manufacturers.

The Simatic MV420 1-D/2-D code reading system with 640 × 480-pixel resolution has an integrated M12 lens and supports operating distances from 1–22.5 cm. Operating distance is adjustable using an integrated focus. An external lighting unit may be connected. The device processes up to 40 codes/s. Using multicode reading, several codes and different code types can be decoded from a single image. The reader features Profinet IO, Industrial Ethernet, and RS-232 interfaces.