AIA Logo

Member Since 1984


AIA - Advancing Vision + Imaging has transformed into the Association for Advancing Automation, the leading global automation trade association of the vision + imaging, robotics, motion control, and industrial AI industries.

Content Filed Under:



A Look at Smart Cameras and Vision Sensors

POSTED 05/19/2015

 | By: Winn Hardin, Contributing Editor

ersity is a hallmark of a strong portfolio, and that doesn’t just apply to financial assets. It’s also the sign of a healthy machine vision industry.

In the 1970s and ’80s, practically every machine vision system — and there weren’t many — used a Versa Module Europa (VME) backplane to connect microchips that formed the system’s brain. By the 1990s, personal computers and the first smart cameras expanded the number of architectures engineers could use to develop a machine vision solution. Today, the smart camera segment has split again into vision sensors and smart cameras, giving customers even more control over the price and power of their machine vision solutions.

While the line separating smart cameras and vision sensors is solidly gray, changing based on who you ask, there are some areas of common agreement that can help customers decide which solution is best based on their application.   

“Cognex kind of invented the vision sensor when we came out with Checker,” says Rick Roszkowski, senior director of marketing, Vision Systems Business Unit at Cognex (Natice, Massachusetts). “What we’ve seen since then is an evolution that goes both upstream and downstream from what Checker can do. Our view is that vision sensors are really geared toward simple functions: pass-fail, go, no-go…with the primary theme being that they’re easy to commission and don’t require a vision engineer to deploy. We see the competition coming in with vision sensors that have both more and less than Checker, but less capability than our In-Sight smart cameras. These systems tend to be   absence/presence verifying and more application-specific.”

“Easy to commission” is one of the commonly accepted elements of the vision sensor definition. “Often customers say ‘smart cameras’, but we call them vision sensors,” says Michael Steinicke, product manager in the Vision Competence Center at Baumer (Radeberg, Germany). “The key point for us is usability when it comes to our vision sensors. We think of a vision sensor as a vision system that only requires configuration, not programming.”

Baumer has developed a line of entry-level vision sensors specifically designed for code reading and check-and-sort applications. Baumer’s VeriSens entire line of vision sensors with resolutions from VGA to 2 megapixels (MP) and on-board FPGA acceleration offers a total library of 21 image-processing functions for its smart sensor line, but two of its five vision sensors only use a subset of the 21 features. The VeriSens ID series, with optical character recognition (OCR) and verification (OCV) functionality, also can locate and read 1D and 2D machine codes. Meanwhile, the CS series is designed specifically for check-and-sort applications. The higher-end XF series also supports pick-and-place operations, providing not only absence/presence but also offset data for robotic and motion control guidance.

Baumer recently introduced the XF IP 69K VeriSens model optimized for both code reading and pick-and-place that comes in a stainless steel housing for applications in food and beverage and pharmaceuticals. The XF and XC (color version) use all 21 image-processing features along with integrated lighting control. To help new customers succeed in color applications, Baumer has developed a 3D color visualization tool based on LAB color space.

Getting More Insight Into Your Application
The latest additions to Cognex’s In-Sight smart camera line illustrate the growing power of the smart camera system despite thermal load limitations inherent to a complete vision system in a passively cooled housing — an important criteria for smart cameras that depend on simplicity, reliability, and small form factor.

Cognex’s new In-Sight 5705 color and monochrome models and In-Sight Micro 8405 offer 5 MP sensors. To handle the additional data-processing demands without sacrificing line throughput, Cognex uses a dual-core DSP/ARM processing element and has spent the past several years optimizing its algorithms for the new In-Sight processing cores and DSP architecture.

“These new models run significantly faster than any previous In-Sight models, in particular the PatMax RedLine pattern-matching tool,” says Roszkowski. “PatMax RedLine runs between 4 and 12 times faster than our previous PatMax running on In-Sight cameras. This has allowed us to leverage the larger 5 MP sensors without taking a hit on throughput at the production line.”

Cognex also has revamped many of the image-processing filters to run faster on the latest In-Sight camera. This functionality is contained in the latest In-Sight Explorer 5.0, which only runs on the latest 5 MP cameras. Finally, the Micro 8405 is highly compact and cost-effective for retrofit operations that do not have high-speed production lines. It uses a rolling shutter CMOS 5 MP sensor to keep costs down while still offering the image processing enhancements of the new 5.0 release.

“The move to higher-resolution 5 MP sensors is in response to customers in packaging, pharmaceuticals, and related industries who now want to use smart camera to package codes up to the case level in a single image,” Roszkowski says. “This new dual-core DSP/ARM architecture and image tool enhancements are the building blocks Cognex needs to eventually go with even higher-resolution sensors on the In-Sight smart camera line. It’s a strategic shift, and certainly non-trivial work to optimize algorithms for DSP/ARM cores. Now, we can use one core for communication and one for image processing and maximize performance.”

Vision in Life Sciences This content is part of the Vision in Life Sciences curated collection. To learn more about Vision in Life Sciences, click here.