Machine Vision

Machine vision technologies leverage digital video imaging and analysis systems to examine and/or inspect materials, objects, processes, incidents, and provide real time processing and analysis for appropriate action to be taken. Typical implementations of machine vision are in conjunction with robotic systems to provide a form of systems intelligence rather than one dimensional behavior. The added smarts brought to robotics have tremendous value in manufacturing processes because previous uni-functional behaviors can now be replaced with intelligent multi-functional responses.

Robust machine vision systems take inputs from a variety of sources/sensors to provide extensive data for appropriate machine response. For example, cameras can examine elements in extremely high resolution and at high rates of speed as well in various visual spectra. In addition, inputs can be taken from other measures, e.g., weight, temperature, etc.  All this data can be fed into the machine vision system and processed and analyzed in real time, resulting in optimized decision making. One application may involve cutting wood to minimize waste. Factors in that process such as the grain and knots as well as the size and shape of the lumber can all be addressed by a single system. These analytical measures can direct precise cutting in terms of depth, angle, shape, etc., and optimize both the process and efficiency, thus adding to the bottom line.

The benefits of high speed and high visual acuity are best when combined with visual processing and analysis technologies. Therefore collection devices are just a single element while the heart and brains of the machine vision solution are the intelligent platform itself. Collecting data from various sources, analyzing the information, and integrating with other systems for appropriate actions, all at extremely high speeds with meticulous reliability is a requirement.

Creating such a platform of machine logic is a formidable technical challenge. Whether systems are deployed in high technology clean rooms or in extreme temperature "dirty" manufacturing floors, the requirement for high reliability and fault tolerance is critical. Appropriate attention must be paid to physical environmental demands to guarantee very low Mean Time before Failure (MTBF) rates.

For example, Boulder Imaging developed an automated quality inspection system for Lexmark that does much more than just “pass” or “fail” a printer drum based on the presence or absence of surface defects.  The quality of the surface of a printer cartridge drum is directly tied to the quality of the printer page. Therefore, it's critically important to have a very uniform surface otherwise field defects, field non-uniformities or any artifacts show up on the printed page.  The system developed for Lexmark included a linescan-based system that both inspects the drums at rates of 80 parts/min and guides robots to sort the drums into “pass” or “fail” groups. In addition, an operator workstation for each line allows an engineer to view defects as well as obtain a report, in real time, that precisely pinpoints where in the production process the defect was introduced – on the raw material or during any of the coating processes.  As a result, problems in any part of the production process can be addressed in real time.  This rule-based classification into likely causes on the production line vastly improved overall assembly line quality and control.

The defects can be extremely small (bubbles, scratches, etc down to 50 microns) and require powerful image processing to classify. The BI system reduced costs by 40% due to defects in the printer drum manufacturing process.

The BI system can also easily grow with the customer. As Lexmark creates new models, a simple tuning of algorithms for defect detection is all that is required. In effect, this solution addresses both current requirements and future needs while cutting down historical problems with false positives.

Boulder Imaging also had to address environmental and materials challenges in the solution design and implementation. Factors such as surface reflection, lighting, camera selection, etc., were all carefully integrated to create this successful machine vision implementation.

Boulder Imaging Meeting Your Machine Vision Needs

Boulder Imaging has implemented a range of machine vision systems to ensure products comply to manufacturers specifications. We have developed systems to ensure appropriate performance in multiple and ranging, and variable conditions. Combining Boulder Imaging’s expertise in high speed, high performance digital recording (HPDVR), processing and analysis systems and machine vision provides the highest quality manufacturing process intelligence. In addition, Boulder Imaging can integrate non-visual data in machine vision challenges as well, resulting in unsurpassed analysis, automated decision making, quality improvements, and process efficiencies.

Boulder Imaging has extensive experience in implementing machine vision solutions in harsh environments (extreme G forces, continual lighting variances, varying temperature exposures, etc.). We make sure that our systems address not only the direct, but also the larger challenges of systems integration, expandability, adaptability, fault tolerance, and much more. To learn more about Machine Vision technologies and how Boulder Imaging can assist your organization in addressing particular needs, please contact us.