Vision systems increase productivity
Vision systems now analyze physical processes for control and automation
By Steve Maves
Straight line illusion
Advances in machine vision (MV) technology are creating new opportunities to increase productivity, quality, and efficiency in a wide range of applications. MV uses computer cameras in manufacturing to capture and analyze images by extracting data and assigning properties to the features of the target item. In the past, vision systems were expensive and difficult to use, but today, these systems are cost effective, have more capabilities, and are easy to apply thanks to advances in software. MV encompasses a large number of related technologies, hardware, and software products integrated into a system solution. MV continues to be applied in a number of creative ways in a range of industries. The main categories into which MV applications fall are quality assurance, sorting, material handling, robot guidance, and controlling production.
Visions systems have a great deal of advantages including elimination of human vision bias. Consider the example of the circles and the square box, where to the human eye the lines of the box do not appear to be straight. In fact, they are straight, and a vision system will objectively detect this.
MV can provide precise inspections at high speeds verifying attributes and guide products through a process. The value includes removing human inconsistency, increased productivity, higher production throughput, improved cycle time, higher quality, and scrap reduction.
Technology improvements such as more powerful computer processors, high-speed cameras, and icon-based software are making machine vision systems easier to apply. In addition, bandwidth is increasing to allow faster image capture.
Initially MV systems were totally custom designs requiring a wide range of specialized skills. Today, icon-based programming environments that require no programming expertise result in short development cycles, higher performance, and flexibility. The latest smart machine vision cameras incorporate a processor providing a complete vision inspection solution.
A simple high-level view of how MV operates is with a trigger that signals to the camera that an object to be inspected is in place; the system takes a picture and enhances the captured image. The system then extracts the desired information from the image and makes decisions based on user-defined parameters to provide an output to machine controls and manufacturing systems. Another major innovation is cameras that communicate using Power over Ethernet (PoE) that also supplies power for the camera over the communications link.
Smart camera elements
An external light source, typically LED, is used to illuminate the target. There is a camera lens, and inside the camera is a digital imaging chip, called an image sensor, and a camera “brain” or CPU that enhances the image, analyzes the image extracting information and making judgments, and finally outputs the results. Each pixel of the image is analyzed based on light absorbed on the image sensor; for example, with an 8-bit camera, the absorbed light for each pixel will have a value of 0 to 255.
For example, a smart camera system operating on a conveyer system where there is a proximity sensor to trigger the camera, at which point the camera captures an image of the target and makes a decision if a part is defective and sends a signal to an actuator to reject the bad part.
Software improvements are making it easier to apply MV and provide more useful displays to operators. Users now work with icon-based software and use drag-and-drop programming with standardized tools including measurement, surface flaw, code reading, guidance, logic, or communication tasks.
Most software suites include an emulator to create, test, and debug programs, without a camera. The object-oriented software allows users to create reusable control panel templates for quick development. Inspection tool settings can typically be edited while inspection is online.
Multiple control panels can connect to the same camera to create unique views of the application for operator, quality control, engineering, management, and others.
Using multiple cameras that communicate over Ethernet with inter camera communications, multiple images of a target can be captured for more detailed analysis. For example, three sides of simple box can be viewed at the same time.
Bar code reading
MV cameras can also read 1D and 2D bar codes without the perfect alignment required with dedicated bar code readers. 2D bar codes can hold a large amount of information in a small space. Human readable information such as part numbers or date codes printed as text on a product can also be read using optical character recognition.
Industrial control, communications
The action signal from an MV system’s analysis and results can simply be digital I/O bits or more significant data bytes and even images communicating critical information to a network. Information sent can include the judgment information from vision analysis and images captured for product genealogy, analysis, archival, and to meet other requirements.
Tips for using MV
MV is an engineered solution, and attention to details will yield excellent results. MV can accomplish a number of things, so it is important to manage scope creep by developing a firm specification describing the specific problems to be solved. Define required function and features, and separate those from the function and features desirable in order to know what can be eliminated if budget becomes an issue.
Qualify and calculate integration costs including internal engineering and a professional MV integrator if appropriate. Training is also important to get the most benefit from MV.
Food production application
Food manufacturers are held to strict agency regulations regarding product quality and consistency. Repeatable visual inspections are critical to ensure consumers receive a consistent and safe product. As regulations have increased, MV technology has become more intelligent to follow suit, and the industry is leveraging MV to improve accuracy and efficiency.
The Budé Group, a company in Masstricht, Netherlands, supports customers such as food manufacturers, designing and manufacturing equipment and even full production lines for diverse applications. With in-house capabilities including controls design, programmable logic controller, and robot programming already employed, Budé extended their company’s core technologies portfolio by developing machinery with integrated high-level MV capabilities.
The Budé Group uses smart cameras and MV systems in food processing and food packaging. They deploy the vision systems to inspect for food and food packaging quality, verify correct product shape and size, and verify correct package markings. Smart cameras also guide food products through the production process and identify barcodes and human-readable text on food packages. Vision system cameras, such as the one Budé selected for their food industry equipment, can help assure end users in the food industry that only the highest-quality products will reach consumers. Using vision, the machines are flexible for handling a range of product types and package sizes. The entire system meets hygienic requirements of the food industry.
The system incorporates up to four smart MV cameras, each capable of performing its own independent inspection task utilizing a single processor, reducing cost per inspection point. These inspections can be initiated independently, at different times or simultaneously, increasing application flexibility, yet the shared processor allows communication between all cameras. The vision system utilizes icon-based vision development and display software. The software suite is used to develop custom graphical user interfaces and create complex inspections via dragging-and-dropping icons for simplified programming. Each camera only requires a single cable connection since they support PoE, eliminating local power requirements at the camera.
Video flame monitoring
Digital video technology is being used for continuous monitoring of the burner flames of furnaces, heaters, and boilers to provide information and feedback on tube temperatures and burner parameters. Traditionally, the operation of fired heaters has been monitored and adjusted by experienced engineers and operators who have looked through doors on the side of the furnace. This brief periodic viewing is to determine if the current operation is acceptable, and based on experience, they adjust various parameters to make the furnace operate safer and efficiently. Vision-based furnace monitoring provides a continuous automated system to operate the burners within parameters, to prevent damage or breakage. Small leaks or pinhole damage can also be detected as they occur. Alert and warning information is sent to operations and the control room for an immediate response to a burner or burners that have drifted outside its preselected parameters.
Vision-based furnace monitoring gives detailed information on what specific burner or burners need to be adjusted for proper air and fuel mixture. With these adjustments, the furnace will be optimized, resulting in an equal heat flux throughout the radiant box, allowing air flow and combustion to be minimized and throughput of the furnace to be maximized. Linked with a historian database, key pieces of data and images are available for examination and analysis of patterns, and problems are archived for analysis. Progressive modeling of the images can then be used to demonstrate and train plant personnel about conditions or procedures that could potentially be harmful. In the situation where an event does occur, the historian files allow the process engineer to retrieve and review data and visual information for analysis. In addition, with remote Internet connectivity, diagnostics, and analysis of potential problems can be viewed remotely so experts can examine and analyze potential problems.
Vision-based monitoring provides the means for a furnace to be operated in a safe and cost-effective way. The monitoring system allows the burners to be tuned to the design excess air, providing an even heat distribution throughout the radiant box without jeopardizing safety of the operation, yet assuring correct air flow and combustion.
Vision systems are being used now in a wide range of applications beyond simply “watching” a process. Coupled with software, they provide automation engineers with a wide range of new tools for monitoring, control, and analysis.
ABOUT THE AUTHOR
Steve Maves is Applications Engineering Manager at PPT Vision (www.pptvision.com).