IMAGING Product News: The Swiss knife of SD/HD acquisition ... · Design Assistant’s simplified fixturing automatically positions (i .e ., returns the x-y coordinate and angle for)

Case Study: Smart camera powers vision system of award-winning food printer

The Vision Squad Files:Pattern matching revisited

Product News: The Swiss knife of SD/HD acquisition

Vol. 11 No. 1Solutions, products and news from Matrox Imaging

IMAGINGINSIGHT

ContentsIMAGING INSIGHT Vol. 11 No. 1

VP’s Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 03François Bertrand, Vice President, Sales & Marketing, talks about what a difference a year makes Hardware Product News . . . . . . . . . . . . . . . . . . . . . . . . . . . . 04Matrox Orion HD: the Swiss knife of SD/HD acquisition

Software Product News . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 06Matrox Design Assistant 2 .3 adds bead inspection tool, simplified fixturing, and Windows® 7 support

Case Study: De Grood Innovations . . . . . . . . . . . . . . . . 07Smart camera powers vision system of award-winning food printer

Case Study: MAF Roda Agrobotic . . . . . . . . . . . . . . . . . . 11MIL ensures only the highest quality fruits and vegetables reach grocery store shelves

Spotlight on Imaging Applications . . . . . . . . . . . . . . . . 14Department receives 500 to 1,000 requests for technical support services and consultations each week

The Vision Squad Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Pattern matching revisited

11

06

07

02 IMAGING INSIGHT Vol. 11 No. 1

17

A year ago, in the depths of a recession, Matrox Imaging decided to move ahead with plans to launch a highly anticipated Authorized Integrator program . We knew that the creation of the new channel would require significant investments in terms of financial and human resources . All of this took place at a time of when many companies’ first instincts would have been to put major initiatives on hold and retreat to protect their existing customer base .

It’s now clear that we made the right decision to soldier through those tough economic conditions and launch our network of vision integrators . Today, our traditional OEM business is stronger than ever and the new channel has helped us develop relationships with some of the world’s largest and most innovative food and beverage, packaging, and automotive companies .

Despite this quest to expand into new markets with products like the Matrox Iris GT smart camera and software bundle, we have not forgotten our roots as a leading-edge designer of board-level products . In this issue of Imaging Insight, we introduce you to our latest frame grabber, Matrox Orion HD, a high-performance graphics adapter that meets demands of both SD and HD video acquisition .

Also, in the magazine, we turn our attention to the food processing industry where we profile two very interesting applications . MAF Roda Agrobotic uses the Matrox Imaging Library to ensure that only the best fruits and vegetables reach our dinner table, while De Grood Innovations worked with one of our integrators to develop a smart camera-based food jet printer .

We hope you enjoy this issue of Imaging Insight .

François BertrandVice President, Sales & Marketing

What a difference a year makes

From the VP’s Desk

Corporate headquarters:Canada and U.S.A. Matrox Electronic Systems Ltd.1055 St . Regis Blvd .Dorval, QC H9P 2T4CanadaTel: +1 (514) 685-2630 Fax: +1 (514) 822-6273

GermanyMatrox Electronic Systems GmbH Inselkammerstr . 8 D-82008 Unterhaching Germany Tel: +49 (0) 89 / 62170 0 Fax: +49 (0) 89 / 614 97 43

Offices:Europe, Middle East & AfricaMatrox VITE LimitedChaplin HouseWidewater PlaceMoorhall RoadHarefieldMiddlesex United Kingdom, UB9 6NSTel: +44 (0) 1895 827300Fax: +44 (0) 1895 827301

IMAGINGINSIGHT

Want to subs cribe? Go to:www .matroximaging .com/newsletter

Publisher: Matrox ImagingEditor / Marketing Communications: Catherine OverburyGraphic Designer: JF Ruest

Reproduction in whole or in part, without the prior written permission of Matrox Imaging, is prohibited . For more information on articles published in this issue, or to share your comments, please contact the editor:

Tel: 1-514-685-7230 ext. 2459 Fax: 1-514-822-6273Email: [email protected]

Commentary

03IMAGING INSIGHT Vol. 11 No. 1

04

Product News: Hardware

Matrox Imaging system host boards (SHBs), utilizing the latest Intel® Xeon® processor, 5600 series (formerly codenamed Westmere-EP), are now available for our high-performance computing (HPC) platform, Matrox Supersight e2 . Intel Xeon processor 5600 series is based on the latest generation Intel® microarchitecture (codenamed Nehalem) using a new 32 nm process technology . The Intel Xeon processor 5600 series is available in both quad-core and hex-core (six-core) versions . In addition, the Intel Xeon processor 5600 series includes a 12 MB shared L3 cache and connects to DDR3L-SDRAM running at up to 1333 MHz . This provides higher bandwidth and better memory performance for I/O-bound operations — a significant benefit for imaging applications .

The term high definition (HD) video has traditionally been associated with the consumer electronics and broadcast industries (e .g ., HDTV) . With its higher resolutions, standard high definition video is increasingly used by industrial markets where image quality is as important as it is for TV broadcasting, if not more . In fact, HD video is now favored over its standard definition counterpart in applications like endoscopy and flight training simulators . Furthermore, the ability to display HD video with minimal latency is as critical to these applications as the capture of HD video itself .

For the imaging market, this represents a shift from basic standard definition (SD) analog acquisition cards to high performance cards capable of capturing and displaying multiple HD video sources . Matrox Imaging’s newest board, Matrox Orion HD, is a graphics adapter that addresses the needs of this market by supporting standard and high definition video inputs and outputs up to 1080p/60 .

Orion HD captures from two independent digital or analog HD video sources and can output these video streams, which can be different formats, on the desktop . With support for different interfaces and flexible connectivity, the board can also switch between SDI, DVI-D, RGB, CVBS (up to 3) and YPrPb signals on each independent path . This allows for capturing and displaying different formats: SD (NTSC and PAL), HD (720p, 1080i and 1080p), or typical PC resolutions up to 1920x1200 .

With its Matrox-designed graphics controller, Orion HD acts as a primary or secondary display device . Captured video and images are displayed on the desktop or to a separate monitor exclusive to the application . Also supported is the option of adding standard graphics annotations .

To accommodate the demands of archiving and playing back multiple HD video streams, Orion HD incorporates a sizable amount of on-board buffering and a PCIe® x16 host interface . This graphics adapter is fully supported by the Matrox Imaging Library (MIL) application development toolkit under 32/64-bit Microsoft® Windows® 7 .

For learn more about Matrox Orion HD, please contact your local Matrox Imaging representative .

Matrox Supersight e2 HPC platform now available with the Intel® Xeon® processor 5600 series

Matrox Orion HD: the Swiss knife of SD/HD acquisition

Hardware News

IMAGING INSIGHT Vol. 11 No. 1

About Matrox Supersight e2 Matrox Supersight e2 is a high-performance computing platform designed for imaging applications with an extremely high throughput . Leveraging multiple clusters of CPUs, GPUs and FPGAs, Matrox Supersight e2 provides an environment for considerable data and task-level parallel processing through a unique PCI Express® (PCIe®) x16 2 .0 (Gen2) switch fabric . Applications for Matrox Supersight e2 are developed with the Matrox Imaging Library (MIL) and its Distributed MIL API . As MIL is supported on all Matrox Imaging hardware platforms, development on Matrox Supersight e2 is simplified and ensures that the source code is portable across all devices, from nodes with 48 CPU cores and 4 FPGAs, to a node with 12 CPU cores, 1 FPGA and 6 GPUs, and everything in between .

05

Product News: Hardware


The Matrox Iris GT for Matrox Imaging Library (MIL) smart camera line is designed for system integrators, machine builders and OEMs who need to deploy fully custom applications . Running either Microsoft® Windows® Embedded CE or Windows® XP Embedded, and offering keyboard, video, and mouse (KVM) support, these cameras provide a PC-like environment for applications . Application development is done using standard Windows® development tools in conjunction with the Matrox Imaging Library (MIL), a software development kit with interactive software and programming functions for image capture, processing, analysis, annotation, display, and archiving .

All Matrox Iris GT smart cameras offer a dust-proof, immersion-resistant and extremely rugged construction . A choice of image sensors, combined with an efficient Intel® Atom® embedded processor, allow the cameras to tackle a wide variety of machine vision applications . The developer chooses whether to work with Microsoft® Windows® Embedded CE or Windows® XP Embedded (XPe) .

Hardware News

Matrox Iris GT smart camera now runs Microsoft® Windows® XP Embedded (XPe)

About Windows® Embedded CE and Windows® XP Professional (XPe)

Windows® Embedded CE delivers real-time performance in a smaller memory footprint and is programmed using a subset of the Windows® API . Matrox Iris GT running Windows® Embedded CE includes support for the EtherNet/IP™ and MODBUS® industrial communication protocols . Windows® XP Embedded is a pared down version of Windows® XP Professional that provides extended availability and support . For both operating systems, application development is cross-platform — PC linked to smart camera — and is done using the familiar Microsoft® Visual Studio® integrated development environment .

06

Product News: Software


Matrox Design Assistant 2 .3 is the latest release of the integrated development environment (IDE) for the Matrox Iris GT smart camera . The software offers many new features and enhancements including a bead inspection tool and simplified fixturing . As well, the Design Assistant 2 .3 IDE can now be installed and used on 32/64-bit Windows® 7 . This release is a follow-on to Design Assistant 2 .2, which enabled the use of the built-in keyboard, video (monitor), and mouse (KVM) capabilities .

Bead inspectionDesign Assistant 2 .3’s bead tool inspects material applied as a continuous sinuous bead, such as adhesives and sealants, and identifies discrepancies in application length, placement and width, as well as discontinuities . The allowable bead width, offset, gap, and overall acceptance measure can be adjusted to meet specific inspection criteria .

Fixturing made simpleFor applications where mechanical fixturing of parts is impractical, Design Assistant’s simplified fixturing automatically positions (i .e ., returns the x-y coordinate and angle for) the region of interest (ROI) for a measurement or reading tool using the results from a locating tool .

Additional enhancements This latest version of Design Assistant offers improved project switching capabilities, support for MODBUS® master mode for controlling automation devices, and simplified flowchart branching using the switch-case element .

Matrox Design Assistant 2 .3 will be available in Early Access form by end-Q2 2010 . Its official release is scheduled for end-Q3 2010 .

New release of Matrox Design Assistant smart camera software offers bead inspection tool and simplified fixturing

Figure 3 – The new simplified fixturing in Matrox Design Assistant 2 .3 automatically positions (i .e ., returns the x-y coordinate and angle for) the region of interest of reading and measurement steps (e .g ., Blob Analysis step) using the results from a locating tool (e .g ., Model Finder step) .

Figure 1 – Matrox Design Assistant 2 .3 includes a new Bead Inspection step to inspect material applied as a continuous sinuous bead for length, placement, width, and discontinuities .

Figure 2 – Configuring the new Bead Inspection step in Matrox Design Assistant 2 .3 . The tool automatically and optimally places the search boxes based on a user-supplied coarse path .

Software News

Decoration goes digital

07

De Grood Innovations (http://www.foodjet.nl) (Nijmegen, Netherlands) is a family-owned and -operated business that specializes in machining and constructing stainless steel, ferrous and non-ferrous metals, and plastics into parts for the medical, industrial, and food industries. Shortly after joining the business 4 years ago, Pascal de Grood, managing director, attended a party where he spoke with a guest who was looking for a printer that could print chocolate on pastries.

De Grood had previously worked for a Dutch research institute and developed printer heads for industrial applications. With his knowledge of printing, he saw possibilities in designing a printer for food manufacturers and began working on a prototype that could digitally print custom-designed, edible high-viscosity decorations onto mass-produced food products. This food printer would also need to quickly switch from one decoration/design to another and from one food product to another with minimal production downtime.

Matrox Imaging smart camera powers vision system of award-winning food printer


Case Study: De Grood Innovations

Entrepreneurial spiritInitially, the design work was done in de Grood’s spare time, spending a year testing and trying to develop the printer head . Once he designed a printer prototype that worked well, he spent the next two years optimizing the system . During that time, De Grood received patents for the technology in Europe and North America .

The printer’s development work was entirely funded by De Grood Innovations . Pascal de Grood explains, “We invested 3 years and a lot of money in this technology, which was not an easy feat for a small business . But we managed, and our first machines generated a lot of interest from potential customers .”

Because of its low maintenance requirement, ease of cleaning, and low price per droplet, the De Grood Innovations BV FoodJet printer is ideally suited for the food industry . Main target markets are bakeries, and manufacturers of ice cream and dairy products .

The FoodJet printer is designed to accurately deposit a variety of thick food materials such as frosting, yogurt or sauce onto food substrates . The system’s series of pneumatic membrane nozzle jets deposit small drops onto the moving food products . These drops then form a digital image in the shape of a decoration or a surface fill . Decorating requires higher resolution and smaller nozzle (droplet) size while filling can be done using big drops and lower resolution .

Configurable printing systemThe FoodJet printer has a flexible, modular architecture that can be configured to any production process . This allows for a vast number of decorating and filling options . Changing the angle between the printer head and movement direction of the conveyor allows the resolution to be set . Using heads behind one another permits “multi-color” printing, where colors are applied in different layers on top of each other . Placing heads next to each other allows for a wider working area .

While the physical placement of the printer head offers a lot of possible configurations, other variables like nozzles, pressure and dispensing time also increase the printing possibilities .

The printer is controlled by two PLCs, and the mechanism to program the various dispensing-pattern options is a homegrown solution, explains de Grood . Sensors like proximity switches and photocells are used for alignment . Depending on resolution, the speed of the printer can be up to 30 m/min with resolution as high as 30 dpi . The system has a reaction time of 200 µs . And with camera, the system accuracy is better than +/-1 mm . The HMI allows the operator to choose product, pressure, and shutter times of valves . However, the camera runs autonomously .

Overnight modeTypical for the food industry, many products need to be manufactured continuously, and, once a process stops, the machine needs to be shut down and cleaned because products cool down and harden or dry . “The FoodJet printer is a closed system, which prevents the product from drying or cooling,” explains de Grood . “This means that it has no problem standing idle for hours or even overnight .”

The traditional food decorating method is manual application or, if the process is automated, using masks and mechanical restraints . Area filling is done by applying a liquid curtain under which the food products are moved on a mesh conveyor . Another filling method involves partially submerging the products in a bath to apply the substance .

The baths and curtain processes especially require an extensive cleaning procedure when switching between products—for example, when switching to a different color frosting or switching from milk chocolate to dark chocolate . Because the FoodJet printer is a closed system, the switch can be done by connecting the reservoir with the new product and flushing the system before connecting the return flow to the reservoir . “A closed system prevents mixing of product ingredients or environmental contamination to ensure the best possible hygiene and food safety,” says De Grood . “One of our



De Grood Innovations FoodJet printer The printer’s vision system uses a Matrox Iris GT smart camera (1) . Additional reservoirs contain liquid that can be used within minutes if a switch is needed (2) . The HMI provides the operator with information on choice of product, pressure, and shutter times of valves (3) . Printing heads (4) . Source: De Grood Innovations

Operator viewThe Dutch-language operator screen reveals fondant supply (aanvoer) and pump activation . The HMI also allows the operator to choose product, pressure, and shutter times of valves . Source: De Grood Innovations


customers managed to reduce the time it took him to switch between decoration fluids by 90 %, as he no longer needed to spend four or five hours cleaning his machine . The transition time with the FoodJet printer now only takes several minutes .”

Machine visionThe addition of a machine vision option to the FoodJet printer came about after a customer approached De Grood with an application where it was virtually impossible to mechanically align the food products so that they could be decorated properly .

“A lot of bakery products would be almost impossible to handle by the FoodJet printer if it did not offer a vision option,” explains de Grood . “These include fragile products which get damaged if handled too much by mechanical means for alignment and products with non-uniform shapes . Both of these situations can be solved by the vision system which measures the product’s exact position and dimension .”

Another advantage of using a vision system is flexibility . The FoodJet printer with the vision option allows for ultra-fast switching between food products . There is no need for food manufacturers to change any mechanical alignment tools or other sensors to work with a larger or differently shaped product .

The FoodJet’s vision system is based on a Matrox Iris GT smart camera . The application was developed with Matrox Design Assistant, an integrated development environment that is bundled with the camera . The software allowed De Grood to create a flow chart of the application instead of coding programs or scripts . This eliminates the need to program in any standard programming language like Visual Basic, C, C++ or C# .

A number of Design Assistant tools or flow-chart steps were used . The location and size of products need to be measured, so calibration of the system is necessary . Image processing filters, blob analysis and model finder—geometric pattern recognition—steps were also used to get the required results . “The flow chart is configured and tested in Design Assistant, an interactive design utility that runs on a PC, and then uploaded to the camera,” explains de Grood .

With Matrox Design Assistant, an application is created by constructing a flow chart using ready-made or custom tools instead of writing traditional program code, explains Fabio Perelli, Smart Camera Product Manager for Matrox Imaging . “Once development is complete, the project, or flow chart, is uploaded and stored locally on the Matrox Iris GT smart camera,” he says . “The project is then executed on the smart camera independent of any PC and, in this case, is monitored and controlled from the PLC .”

DVC, the Dutch distributor for Matrox Imaging, developed a custom step for communicating with the PLC . The ability to create custom steps is one of the features of the Matrox Design Assistant environment, enabling users to insert application-specific logic on their own . This step calls an .exe file that DVC made to communicate with the PLC . The timing of the communication between the Iris GT smart camera and PLC was critical . “It took a little while to get the camera and the PLC in sync, but after a day of tweaking, the communication ran smoothly,” says de Grood .

Vast number of decorating and filling optionsChanging the angle between the printer head and the conveyor’s direction of movement changes resolution (at the expense of coverage) .

Vast number of decorating and filling options (continued)Positioning printer heads behind one another permits multi-color printing (1) . Placing heads next to each other allows for greater coverage (2) .



Several smart cameras were evaluated by De Grood for the printer’s vision system but Matrox said the application was possible and the Iris GT’s price was less than other alternatives, explains de Grood .

It also offered the IP67 form factor that de Grood wanted . “The IP67-rated housing was welcomed since this makes for easy integration in the system, without the need for additional protection against moisture,” he says . “To keep the system as compact as possible, we wanted to have the camera and processing combined into one package .” The printers also link back to the enterprise system, offering the possibility of full remote assistance worldwide . A future beyond foodDe Grood emphasizes that the FoodJet printer can be used by industries other than food processing . “Every production method that applies thick, highly viscous fluids is a potential market . Some non-food examples are decorating ceramic tiles, applying complex glue patterns to printed circuit boards, ICs, flat screen components, etc . or printing a company logo on cosmetic creams to show consumers that they have purchased a real product and not a cheap counterfeit . The possibilities are endless .”

De Grood Innovations BV FoodJet printer was nominated for a 2009 Food Valley award . It is a prize awarded each year by Food Valley, a Dutch agro-food organization . The 2009 award was announced October 8, 2009 .The FoodJet printer won second prize .

A condensed version of this article entitled “Food Printer Adds Vision System to Handle Delicate Pastries” appeared in the February 2010 issue of Control Design . Used with permission .


Smart camera-based vision systemThe FoodJet’s vision system is based on a Matrox Iris GT smart camera . The application was developed with Matrox Design Assistant, an integrated development environment bundled with the camera . The software allowed De Grood to create a flow chart of the application instead of coding programs or scripts .

Matrox Imaging Library (MIL) ensures that only the highest quality produce reaches grocery store shelves

A quest for the best

Case Study: MAF Roda Agrobotic


— looking fruits and vegetables, that is!

It is estimated that more than 30 per cent of fruits and vegetables grown for North American consumers are discarded before they reach grocery store shelves because of “cosmetic imperfections” . An apple is too small, a pear is the wrong color or a cucumber is misshapen . Imperfect produce that does manage to reach supermarkets remains

unsold for the most part and ultimately ends up in landfill sites . In this era of skyrocketing food costs and global shortages it is in everyone’s best interests – growers, packers, distributors, retailers, and consumers – to reduce this waste .

12

Case Study: Case Study: MAF Roda Agrobotic


MAF Roda Agrobotic (http://www .maf-roda .com) is a French multi-national that designs and manufactures grading, packing, washing, bagging, and palletization systems for fresh fruit and vegetable packing houses . MAF Roda helps packers improve processing efficiency, while ensuring that produce is the correct size, shape and color, as well as being free from defects .

History In 1962, MAF Roda began manufacturing mechanical graders in the largest fruit growing area in southwest France . In 1979, Philippe Blanc, a graduate of École Nationale d’Ingénieurs (Tarbes, France), along with ten fellow students, produced the first electronic and automatic grading systems to replace those mechanical graders .

Today, MAF Roda has 18 subsidiary companies and 40 agents located in the most abundant food producing regions in the Americas, Europe, Africa, and Asia . The family-owned and -operated business is now overseen by a third generation – Philippe Blanc’s five sons . In 2008, MAF Roda annual sales were 90732K€, making it one of the largest manufacturers of fruit and vegetable handling equipment in the world .

Sizing and sorting fruitMAF Roda designs and manufactures several different types of “sizers” . These are conveyors with 1 to 10 parallel lanes and 1 to 64 outlets that process different types of fruit ranging from cherry tomatoes to melons . The number of lanes determines the processing capacity of the machine – 10 to 15 pieces of fruit per second, per lane – and the number of outlets defines the number of categories that the machine can sort .

One of these systems is the GLOBALSCAN®, a sophisticated, fully automatic vision system that sorts fruit as it moves along a sizer . The system allows packers to accurately grade fruit according to pre-established quality and color criteria to meet market requirements . Processing and packaging costs are reduced as substandard fruit is immediately removed from production batches .

The vision system includes proprietary double sensor CMOS high-resolution cameras (one color and one infrared camera per lane), acquisition boards, and LED lighting . Twenty color pictures and 20 infrared pictures are taken of each piece of fruit as it moves and rotates forward – covering 100 % of its surface . The vision system can perform optical sizing (equatorial or maximum diameter, volume integration), color sorting (eight different criteria), and detection of external defects on the fruit .

The images are then processed using the Matrox Imaging Library (MIL) toolkit . Binarization and blob analysis are used to separate the fruit from its background and to obtain dimensions like Feret diameters and elongation, for example . The HLS (Hue, Luminosity, Saturation) color space conversion and statistical functions are used for color processing . And convolutions and custom-designed algorithms recognize stems and cores, and reject fruit with defects on the skin like black spots or russet .

The system’s intuitive user interface, which includes a 3D display, lets operators adjust quality and sorting parameters in real-time . These parameters can be stored for future use . GLOBALSCAN® gives production managers the ability to analyze the data gathered from the vision system and generate statistics by

Sizing and sorting orangesThe sizing machine lets a packer in Valencia, Spain accurately grade oranges according to pre-established quality and color criteria to meet market requirements. This machine processes 40 million kilograms of oranges per year. Source: MAF Roda Agrobotic


producer, by orchard or field, by species, or by variety of fruit . Reports provide metrics on the batch of fruit that is being processed including size, weight, diameter, color, and quality .

Off-the-shelf versus proprietaryPrior to using MIL, MAF Roda engineers developed their own image processing library in C, including optimizations using MMX/SSE instruction sets . Initially, the company worried that they would sacrifice performance when they started development using an off-the-shelf library like MIL, because these tools are designed for a more general use than MAF Roda’s “homemade” functions .

“We were happy to see that MIL functions are very fast and powerful . We can now spend more time on our specific job using MIL functions instead of wasting time writing our own functions from scratch . Plus, the stability of the Matrox Imaging Library is also very good,” explains Michel Rodière, Electronics Department Manager . “MIL definitely helped our productivity . Compared with what we did before, MIL saves programming time and MIL code is easier to read and understand when several programmers work on the same project .”

Looking forwardMAF Roda began evaluating Matrox Imaging Library (MIL) 8 at the end of 2007 . The first production machine made its debut November 2008 . They have begun the process of moving to MIL 9 to benefit from the native .NET development and 64-bit Windows® support . They are also looking at making use of the color analysis tools available as of MIL 9 .

Case Study: MAF Roda Agrobotic

GLOBALSCAN® vision systemThe vision system includes proprietary double sensor CMOS high-resolution cameras (one color and one infrared camera per lane), acquisition boards, and LED lighting. Source: MAF Roda Agrobotic

Operator control roomOperators adjust quality and sorting parameters in realtime. These parameters can be stored for future use. Source: MAF Roda Agrobotic

Intuitive HMIGLOBALSCAN’s® intuitive user interface includes a 3D display and provides the operator with information on a piece of fruit’s size and color, as well as any external defects . Source: MAF Roda Agrobotic

14

Meet Imaging Applications


Spotlight on Imaging Applications

15



Matrox Imaging’s Applications Department receives 500 to 1,000 requests per week for technical support services and consultations. Read more to discover what drives this team and how they strive to resolve all issues quickly and accurately.

MandateThe primary goal of the Applications department is to provide technical assistance so that customers can use Matrox Imaging hardware and software to its full potential . On a typical day, every support specialist will respond to a dozen or so issues . He usually has much less than an hour to address an issue and move to the next one .

Processing product enhancement queriesAs well as offering support, the Applications team is often approached by customers requesting product fixes . This user feedback is very valuable to Matrox Imaging, as it lets us understand exactly what our customers expect from our hardware and software in terms of quality and performance . When an issue with one of our products is reported, this information is registered by a support specialist in our information tracking system (ITS) . These issues are then sent to our product developers for evaluation .

Priority support for Matrox Imaging Authorized IntegratorsLast year, Matrox Imaging launched a very successful Authorized Integrator Program . The Applications department offers priority support services to this ever-expanding network of integrators . All integrators have direct access to a team of product support specialists who offer personalized, same business day response .

Remote support capabilitiesIn addition to supporting customers who contact us via email and telephone, Matrox Imaging offers the possibility of remote support sessions if an applications specialist deems it necessary . These sessions allow our staff to connect to a user’s development or run-time system to troubleshoot, diagnose and resolve challenging issues quickly – without requiring an on-site visit .

An ounce of prevention....Customers who are knowledgeable about our products and technologies require less support; this is why the Applications department is actively involved in training initiatives . Product support specialists participate in classroom training sessions and help create and deliver custom, on-site training courses tailored to meet users’ specific requirements and development timelines . These training sessions have proven successful in reducing customers’ learning curves so that projects are up and running more quickly .

16



Matrox Imaging Applications at a glance…

Matrox Imaging’s “front line” support group answers developers’ installation questions and provides immediate troubleshooting assistance . Our team of applications specialists guides customers through the design, development, and deployment phase of their projects .

Matrox Imaging offers developers comprehensive training on our software tools, including instructor-led classroom sessions at various locations worldwide . We also have extensive experience developing custom, on-site training courses tailored to meet customers’ specific needs and project schedules .

The Vision Squad Files: Pattern matching


18


Matrox Imaging pattern matching revisitedAlthough pattern recognition is one of the earliest digital image processing tasks, it still represents a fundamental aspect of many contemporary and future machine vision systems . Over time, pattern recognition software tools have evolved to become highly flexible and tolerant to noise, rotation, scale, occlusion, uneven lighting variations, etc . Nevertheless, certain machine vision applications still benefit from the fast and well-known pattern matching technique based on normalized grayscale correlation (NGC) . As a result, the Matrox Imaging Pattern Matching tool is still subject to continuous enhancement in order to address today’s challenges . This Vision Squad File introduces the latest improvements to this tool and provides some tips and tricks on how to use it most efficiently .

New automatic content-based resolution level selectionThe normalized grayscale correlation of a model image and a target image is a computationally intensive task, which greatly depends on the size of both target and model . To speed up the process, the Pattern Matching tool employs a hierarchical search strategy using a series of sub-sampled images, called the resolution pyramid, where the image at each level is half the size of the previous one .

The search starts at a low resolution level, where attempts are made to quickly locate candidates that are then passed to a higher resolution level for local validation and better positional accuracy . The automatic selection of the resolution levels is the task of choosing the first and last level of the resolution pyramid before starting the hierarchical search . By default, the automatic selection of these levels is based only on the size of the model . However, this simple decision may lead to one of the following situations:

1 . An overly aggressive first-level selection when a small prominent feature is enclosed in a large model image

Suppose a very small circular-shaped object is the only meaningful feature in a large model image . Since the model image is very big, the algorithm assumes that it can work on a highly sub-sampled version of the model . However, at this very low resolution, the small feature has almost disappeared and the hierarchical search will not be able to find any candidates as illustrated in Figure 1 .

Figure 1 - At a very low resolution, the small feature has almost disappeared and the hierarchical search cannot find any candidates .

2 . Suboptimal selection of the first level when a large object is enclosed in the model image

The simple criteria based on the size of the model can lead to the situation where a model image could have been sub-sampled more without affecting the robustness of the hierarchical search algorithm . This is the case when the prominent feature occupies most of the model image as illustrated in Figure 2 .

Figure 2 – A prominent feature occupying most of the model image could be more subsampled

As of MIL 9 Processing Pack 2, the Pattern Matching tool offers a new automatic content-based level selection algorithm that performs a content analysis of the model . This allows the first and last level of the hierarchical search to be chosen more “intelligently” . This new automatic mode is even more useful given that the limitation on the size of the model has been removed since MIL 9 . The mode is easily enabled by explicitly setting a tool parameter (M_FIRST_LEVEL control to M_AUTO_CONTENT_BASED) .

• To adapt to the relaxed constraint on the size of the model image, the first and last level can now be manually set to a value in the range [0, MaxLevel] instead of [0, 7] previously . The MaxLevel is a read-only value provided by the module that can be inquired with MpatInquire(…, M_MODEL_MAX_LEVEL, …)

Additional tips:


ResolutionLevel Search stage

Lowest

Highest

7

6

5

4

3

2

1

0

Initial

Final

Figure 4 – Relative speed improvements for different speed optimizations . Results apply to searching a 32x32 model with ±180° of possible rotation in a 1K x 1K image . Benchmarking performed using an Intel® Xeon® X5472 running 64-bit Windows Vista .

X

Search angle tolerance

Y

2.5

4.1

6.5

8.5

0123456789

1 core withSIMD

instructions

2 cores withSIMD

instructions

4 cores withSIMD

instructions

8 cores withSIMD

instructions

Type of speed optimization

Spee

d-up

ver

sus

no o

ptim

izat

ions


New automatic content-based selection of angle tolerance and accuracyWhen the angle mode is activated, there are a few important parameters that define the behavior of the algorithm with regards to the following aspect:

1 . The range of angles to cover in the search operation

Figure 3 – The range of angles to cover in the search operation .

2 . The tolerance and accuracy of the search

At the preprocessing step (MpatPreprocModel), multiple models, each at different orientations, are generated using the tolerance parameter (M_SEARCH_ANGLE_TOLERANCE) for the angle step . The search algorithm then uses these rotated models to find possible occurrences in the target image . If this initial coarse search succeeds, the angle is refined using the accuracy parameter (M_SEARCH_ANGLE_ACCURACY) .

By default, the tolerance parameter is a fixed value (5 degree), which is enough for most situations although it is not optimal . The new automatic mode intelligently chooses the angle tolerance by analyzing the content of the model image . This mode is enabled by explicitly setting a tool parameter (M_SEARCH_ANGLE_TOLERANCE control to M_AUTO) .

If we rotate the model by 5 degrees but the target occurrence is 5 .5 degrees, the misalignment will result in a lower score . The extent of this decrease in score depends on the shape of the model . For example, a circle is not sensitive to rotation while a line segment will almost not correlate at all as a result of rotation . To help determine the angle accuracy to use, the Pattern Matching tool also offers a new parameter (M_ROTATED_MODEL_MINIMUM_SCORE) to set the minimum acceptable score due to the angle range sampling . In this case, the angle step accuracy will automatically be refined during the preprocessing step with respect to the score constraint . By default, this control is set to M_DISABLE, in which case no automatic accuracy is computed . If a value between 0 .0 and 100 .0 is given, it will be used to automatically compute the necessary accuracy . The maximum rotation angle that gives this score will be used to calculate the accuracy .

To prevent missing an occurrence in the target image due to the angular sampling, the minimum score (M_ROTATED_MODEL_MINIMUM_SCORE) should be set to a value higher than the overall acceptance score (MpatSetAcceptance()) . However, the higher the rotated minimum score, the smaller the refined angular search, and the longer the matching .

Speed optimizationAlso as of MIL 9 Processing Pack 2, the Pattern Matching tool can take advantage of a CPU’s multiple cores to accelerate a search by transparently splitting the task between cores (see Figure 4) . Multiple cores will be automatically used when beneficial, which depends on the size of the model, the number of pyramid levels, the search angle range, the size of the target, and the number of models .


©M

atro

x Im

agin

g P

rint

ed in

Can

ada

$IE-

5484

-B

IMAGING Product News: The Swiss knife of SD/HD acquisition ... · Design Assistant’s simplified fixturing automatically positions (i .e ., returns the x-y coordinate and angle for)

Documents