Utilizing Machine Vision Engines for Advanced Quality Control and Object Detection
- Created: Wednesday, 01 June 2011
Advanced object and feature recognition eliminates the need for physical probing or fixed sensors.
Incorporating machine vision engines into an OEM system with automated handling provides a wide range of benefits to the manufacturer. Machine vision engines are self-contained vision systems that include the optics, lighting, image sensor, electronics, and software to enable standalone “intelligent” decision-making using vision software. Machine vision engines are also much more compact than traditional industrial machine vision systems, which makes them more appropriate for integration into OEM systems with limited space.
Most automated handling systems, particularly in the clinical diagnostic industry, utilize a system of sensors and physical probing to determine the presence or absence of a particular object or feature. In many cases, just using a sensor or physical probe can lead to false object identification due to the single point nature of the sensing.
Image-based engines are not necessarily new. Over the past several years, a few manufacturers of automated ID products for reading barcodes have released derivatives of their products for integration inside OEM systems. Many are familiar with those application uses today in airport ticket scanners and gaming/lottery systems. However, there are now OEMs who manufacture automated handling systems that are incorporating the ability of the engines to also run machine vision algorithms for intelligent object identification to go beyond applications that primarily focus on bar code reading.
There have been two primary challenges, however, to incorporating the benefits of machine vision intelligence into OEM systems: usability and cost. While there are a few image engines that can be purchased on the market for reading barcodes, none of them provide the capability for the OEM to easily program them with the specific vision functions for their application. Instead, the OEM is usually forced to utilize third-party integrators to write custom programs that are specific to their application in C or another programming language. As a result, the OEM is faced with tackling a project to integrate machine vision that is time-consuming, inflexible, and costly.
To address these issues, Cognex’s ‘Advantage’ OEM image engines can be connected to and programmed with their industry-proven Windows-based vision software, which has been deployed in thousands of machine vision applications. OEMs can develop and test a particular machine vision application in minutes, compared to what would have taken them months with a third party integrator. In addition, OEMs have complete control over any changes they might need to make, without the risk of time delays or additional cost.
Machine vision can be used for a wide range of tasks. Almost all machine vision applications can usually be categorized under one of four areas: Guidance, Inspection, Gauging, or Identification. With advanced machine vision software, OEMs can now use vision engines to “see” their parts. Machine vision provides the flexibility that can never be accomplished with sensors and physical probing to provide greater accuracy and system throughput.
An example use of vision in an automated clinical diagnostic system would be for the OEM to incorporate vision to determine whether or not a cap is on or off of a test tube. Powerful geometric pattern location tools are able to recognize unique features of the tube, and automatically position accurate measurement tools based on precise edge detection technology. The OEM can set pass/fail limits on the size to verify the correct cap is on, and inserted properly. Another typical vision application is the counting of objects in the field of view. Simply by drawing a box, the OEM can use the advanced counting tools to count features either by shape, size, or shade, and verify position.
The introduction of compact, powerful, and cost-effective vision engines will open up a completely new alternative to machine builders. While previously limited to the more mechanical sensing of individual optoelectronic sensors or actual probing, machine vision engines provide the ability to introduce more sophisticated accept and reject criteria. In addition, vision provides the ability to error-proof an automated handling system, which is often critical due to the wide range of operator skill to be expected in the field.
This technology was done by Cognex Corporation, Natick, MA. For more information, visit http://info.hotims.com/34455-194.