FIVE EDGE INSPECTION TECHNIQUES IN MACHINE VISION TECHNOLOGY

Finding an edge is one of the most critical functions in machine vision systems, whose algorithms comb through the pixels in a digital image in search of lines, arcs and geometric shapes. Software translates this data into edges, which tell machine vision software which areas to focus on and which ones to ignore.

Thus, edge-inspection tools bear a substantial responsibility for the accuracy and efficiency of machine vision systems. The fundamentals of edge inspection tools illustrate some of the core functions of machine vision.

HOW EDGE DETECTION WORKS IN A FACTORY SETTING

Here’s a common edge-inspection scenario based on Cognex’s machine vision software:
A completed piston assembly must be inserted into a V-8 engine block. A machine vision application takes a photograph of the piston assembly and uses machine vision algorithms to identify its edges. Another picture finds the edges within the engine to block that reveal the piston assembly’s installation location.

Edge-inspection tools are configured to direct the machine vision system to focus its attention on specific areas of the piston assembly and engine block while filtering out everything else. This is crucial because computer processors must scan every pixel within an image, which requires processing time and energy. The system runs best if it scans only the required pixels.

In our example, a machine vision system uses edge inspection data to set up a quality-control application that scans images of the piston assembly and engine block for evidence of defects. Once they pass inspection, they proceed down the assembly line to a robot arm that uses edge-inspection data to tell the robot exactly where to place the piston within the engine block.

Operations like this play out in almost infinite variety, given the widespread prevalence of machine vision technology in distribution centers and factory automation.

5 TOOLS FOR ACQUIRING ACCURATE EDGE INSPECTION DATA

Here’s a common edge-inspection scenario based on Cognex’s machine vision software:
A machine vision system sets up a series of parameters to determine if an item being scanned should progress through the production environment or be rerouted to an area for addressing defects. Every item photographed or scanned gets a pass or fail rating.

Edge inspections can be configured to establish tolerances. Any object falling outside these tolerances can be rejected, while everything within the tolerances passes.

To visualize how edge inspections work, imagine a modern-day factory creating reproductions of old-fashioned wagon wheels, which have three principal parts: the outer rim, the spokes and the hub. Edge inspection parameters are critical to using industrial robots to automate the manufacturing process.

These five edge inspection techniques come into play:

• Distance. In a wagon wheel, the distance between spokes, rims and hubs must fall within tight tolerances. Edge inspection tools measure the distance between these components in a scanned image, enabling both quality control and alignment for robotic production.

• Angle. Each spoke of the wagon wheel has to be installed at an exact angle. An angle edge inspection tool gives the robot accurate guidance on spoke alignment.

• Circle diameter. Manufacturing or distribution flaws might deliver the wrong rims or hubs to the robot. A circle diameter edge inspection measures the distance from the center to the edge, creating data for flagging production errors.

• Circle concentricity. The wagon wheel’s rim and hub share the same center, which makes them concentric. A circle concentricity edge inspection helps the robot align the rims and hubs.

• Radius. The radius of each rim and hub provides more data to ensure that the robotic automation gets them into precise alignment.

Manufactured components as simple as a wagon wheel or as complex as a smartphone circuit board all benefit from these kinds of edge inspection applications.

CHOOSING THE RIGHT EDGE INSPECTION TECHNOLOGY

At Cognex, we’ve been perfecting the art and science of machine vision for decades. Our InspectEdge tool is one of the core assets in our machine vision software suite, which was designed to make it easy for anybody to set up a vision application, even if they don’t have advanced certifications or college degrees.

Other tools in our software suite accomplish essential tasks like bead inspection, pattern matching, identification and image processing. Whether you’re running a distribution center or automating a factory environment, these tools will give you an edge in quality control and efficiency.

TO KNOW MORE ABOUT MACHINE VISION SYSTEM PRODUCT DEALER IN MUMBAI INDIA CONTACT MENZEL VISION AND ROBOTICS PVT LTD CONTACT US AT (+ 91) 22 67993158 OR EMAIL US AT INFO@MVRPL.COM

HOW TO SELECT THE CORRECT MACHINE VISION LENS FOR YOUR APPLICATION

 When setting up your automated vision system, the lens may be one of the last components you choose. However, once your system is up and running, your data flows from the lens first. That makes your lens choice one of the most impactful decisions that affect how well your vision system works for you.

Resolution is a priority. A higher resolution lens gives you greater specificity in designing and implementing the most efficient vision solutions.

Don't let the lens be the weak link in your Machine Vision (MV) system. Choosing a great lens tailored to your system's needs can be daunting. To select the ideal lens, one should consider several factors. So, what is the best way to choose the right lens for a machine vision application?

Selecting a Machine Vision Lens Checklist

1. What is the distance between the object to be inspected and the camera, i.e., the Working Distance (WD)? Does the distance affect the focus and focal length of the lens?

2. What is the size of the object? Object size determines the Field of View (FOV).

3 . What resolution is needed? The image sensor, as well as the pixel size, are determined here.

4. Is camera motion or special fixturing required?

5. What are the lighting conditions? Can the lighting be controlled, or is the object luminous or in a bright environment?

6. Is the object or camera moving or stationary? If it is moving, how fast? Motion between the object and camera has shutter speed implications, affecting the light entering the lens and the f-Number.

These variables and more make selecting the proper lens a challenge, but an excellent place to start is with three significant features: type of focusing, iris, and focal length.

Choosing a great lens tailored to your system's needs can be daunting, but we are here to help. Talk to a lens specialist at Computar today and find out how we can assist in selecting the correct lens for you.

TO KNOW MORE ABOUT MACHINE VISION SYSTEM PRODUCT DEALER IN MUMBAI INDIA CONTACT MENZEL VISION AND ROBOTICS PVT LTD CONTACT US AT (+ 91) 22 67993158 OR EMAIL US AT INFO@MVRPL.COM

HOW A VISION SYSTEM WORKS

MARCH, 2022  ARTICLE

The architecture of a vision system is strongly related to the application it is meant to solve. Some systems are “stand-alone” machines designed to solve specific problems (e.g. measurement/identification), while others are integrated into a more complex framework that can include e.g. mechanical actuators, sensors etc. Nevertheless, all vision systems operate are characterized by these fundamental operations:

Image acquisition. The first and most important task of a vision system is to acquire an image, usually by means of light-sensitive sensor. This image can be a traditional 2-D image, or a 3-D points set, or an image sequence. A number of parameters can be configured in this phase, such as image triggering, camera exposure time, lens aperture, lighting geometry, and so on.

Feature extraction. In this phase, specific characteristics can be extrapolated from the image: lines, edges, angles, regions of interest (ROIs), as well as more complex features, such as motion tracking, shapes and textures. Detection/segmentation. at this point of the process, the system must decide which information previously collected will be passed on up the chain for further elaboration.

High-level processing. The input at this point usually consists of a narrow set of data. The purpose of this last step can be to:

• Classify objects or object’s feature in a particular class
• Verify that the input has the specifications required by the model or class
• Measure/estimate/calculate specifics parameters as position or dimensions of object or object’s features

TO KNOW MORE ABOUT MACHINE VISION SYSTEM PRODUCT DEALER IN MUMBAI INDIA CONTACT MENZEL VISION AND ROBOTICS PVT LTD CONTACT US AT (+ 91) 22 67993158 OR EMAIL US AT INFO@MVRPL.COM

ROCK AND ROLL! MACHINE VISION CAMERAS FOR VR IN LIVE CONCERTS

 AUGUST, 2020  ARTICLE

Not long ago, virtual reality was little more than the stuff of science fiction books and movies. Today, virtual reality is not only making inroads in actual high-end science and technology, but also in ways that affect the life of everyday people from interactive gaming and data-driven sports broadcasting, to video conferencing, education & training, and live music & concerts.

Many music and concert lovers may have had the experience of been shoved around in the audience or had a very limited view of their favorite artists as they struggle to get the best possible view of the stage from the audience. Some people prefer to avoid the front crowds and prefer to sit back in the lawn area or back bench seats where the rowdy crowd behavior is minimal. But choosing this option means there is a risk of missing out on the details of the artists performance or the expressions conveyed to listeners within the proximity of eye contact with the performer. Such moments add to the live feel and reality of a concert experience.

In addition, since the quasi-collapse of the music industry where revenues for physical music tumbled drastically between 2001 and 2018, the industry has turned to live music as its main source of income. With several hundreds of live concerts taking place globally each year, it is impossible for fans to be physically present at all times.

Another challenge for live concerts is that they place a natural limit on the number of people who can attend them. This again, has made it harder for music fans to see their favorite artists. To meet these challenges, virtual reality has stepped in and is now playing a key role in bringing live concert experiences from the front rows to the living rooms of fans and audiences worldwide.

Virtual reality concerts are a win-win situation for the music industry and the music fans. In addition to those who pay to actually be present at a concert, the music industry can monetize everyone who couldn't obtain tickets, or who didn’t always feel like going out to see their favorite musicians perform. Virtual reality lets the music industry combine the best of both worlds: the apparent spontaneity and singularity of live music with the reproducibility and accessibility of recorded music.

Camera technology plays a key role in enabling virtual reality in live concerts. This is because the concerts are captured live from various angles using high-end cameras. The live images are then processed in almost real time to enable remotely-located audiences to choose and view the concert from positions of their choice (e.g., viewed from the front rows, viewed from different acompanists such as percussionsts, guitarists or piansts, different views of the crowds, etc.) all while delivering an immersive experience that goes far beyond that provided by a traditional concert DVD.

From a display perspective, like in sports imaging, the horizontal pixel resolution plays an important role in the quality of virtual reality. This resolution can either be actual or interpolated from a higher or lower raw image format.

4K horizontal resolution for VR has been around for quite some time. 4K, also known as Ultra HD, has a pixel resolution of 4,096 x 2,160 pixels. When compressing video streams from 4K to HD-streamable video the images are clearer, sharpner and cleaner. Shooting at such a high resolution gives editors and image processing engineers an opportunity to zoom far into images and reframe without losing information.

Using an 8K horizontal resolution, also known as Full Ultra HD, allows the user to zoom in twice as much and still get a 4K image. However, achieving real 8K horizontal resolution for virtual reality applications is difficult even if the cameras support 8K horizontal resolutions. This is because most virtual reality installations prefer each camera to have an ultra-wide field of view to give viewers a panoramic or hemispherical view while reducing the equipment handling complexities during live concerts.

The only way to achieve such ultra-wide fields of view is by using fish-eye lenses. Combining a fish-eye lens with a camera using a rectangular sensor is only possible by having an image circle that is smaller than the sensor. Today, sensors with 8K horizontal resolution are able to achieve 5324 pixels in real horizontal resolution when paired with a fish-eye lens of 4.3 mm focal length. This helps to achieve an angle of view of 250° with 21 pixels per degree, which is a good number of pixels for high quality image processing and enhancement. Interpolation can then be used to achieve an 8K horizontal screen resolution.

Obviously, the higher the camera resolution, the closer one can get to achieving real 8K horizontal resolution. But it is important to remember that these are live action events. Higher VR resolution is only useful if a camera speed of at least 30 FPS can be maintained. This limits the choice of cameras that can be used for VR applications.

One final requirement for cameras used in virtual reality applications is reliable data transmission at low noise levels over long distances. Concert venues are typically quite large, and cameras may need to be placed at locations far away from the crowds. CXP and optical interfaces (e.g., SFP+) are reliable and well-known interfaces to handle both the dist

QUALITY INSPECTION OF PHARMACEUTICALS USING HIGH SPEED MULTISPECTRAL IMAGING

 AUGUST, 2020  ARTICLE

Pharmaceutical manufacturing is a complex process which mainly deals with the manufacturing of drugs and medicines. Being a fully automated high-speed manufacturing processes, pharmaceutical production is especially challenging. It is subject to strict regulations given by public health authorities.

Defective containers, incorrect or missing medicine, mislabelling, inefficient packaging or decoloring are risks to consumers and thus different stages of the manufacturing process need to be critically inspected. In order to produce safe medicines that minimize consumer risk and succeed in a competitive market at the same time, there is a requirement for highly effective, versatile, and sensitive quality control systems. Optical quality control using camera technology plays an important role to fulfill the challenging inspection tasks in pharmaceutical manufacturing.

Pharmaceutical products come in various forms and packages, the most common being blister packages and tablets. Those consist mainly of three parts: cavity, seal, and the tablet or drug itself. The cavity is made from synthetic material or aluminium and holds the drug.

Cavity and drug are sealed with a synthetic material, aluminium, paper, or soft foil. Though each component is closely monitored prior to packaging, shortcomings still occur during the primary packaging process. Damage to the package or content, including incorrect placement, coloring, or labelling, must be identified and eventually followed by removal of the defective product.

Production numbers are extremely high for most pharmaceutics. Optical quality control systems along with sophisticated machine learning algorithms can handle large numbers, while offering high sensitivity for defect recognition. Using high speed optical control systems, the whole sample can be inspected, which is a major advantage compared to other quality control systems like manual or mechanical inspection which can end up destroying the sample during the inspection process. There are also limitations on the size of the sample that can be handled using mechanical inspection systems.

For many years, inspection of pharmaceutical packages has been carried out with conventional RGB cameras, using only visible features to detect flaws. With the advent of multispectral cameras, one can now move beyond the visible spectrum. Multispectral cameras capture information of multiple discretely positioned spectral bands, including bands outside the visible region.

In addition to visible R-G-B imaging, the additional spectral bands in multispectral imaging can assist in distinguishing different tablets based on their chemical composition, even if they are already enclosed and sealed. Furthermore, the quantity and uniformity of the active pharmaceutical component (APC) in the tablet can be measured.

The possibility to assess the extrinsic and intrinsic properties at the same time has major advantages compared to conventional quality control inspection systems. Extrinsic properties such as package condition, labelling and dosage instructions, and color coding can be inspected using the visible spectrum. Intrinsic properties of medicinal packages such as breakage of pills, fill levels of liquids, foreign objects and quantity of pills can be captured using specific spectral bands – typically in the near infrared (NIR) region.

Multispectral imaging can also be used in applications related to mistaken identities of defects. For example, in parenteral (injectable) drugs, inspection is critical to verify that there are no particles in the parenteral solution. Multispectral imaging can more easily differentiate between bubbles and particles to minimize waste while ensuring the purity of the injectable medicine.

Advanced multispectral imaging also assists in inspecting the chemical composition of pharmaceuticals. Both, intrinsic and extrinsic information can be combined for quality assessment. This allows the producer to have a single quality control setup, which is generally more robust, simpler to operate and to maintain.

Personalized medicine is going to be an important area of pharmaceuticals in the future where medicines would be manufactured based on an individual’s underlying health conditions, reaction to specific chemicals, and effectiveness for a specific patient. Camera technology combined with artificial intelligence will continue to play an important role in the quality inspection of personalized medicines.

To support high throughput in pharmaceutical production lines, modern inspection systems will need to be equipped with high speed multispectral cameras, which include the ability to inspect multiple spectral bands at high speeds simultaneously. High performance interfaces such as 10GBASE-T (10 GigE Vision) not only have the bandwidth for high frame rates but also support multi-stream output over a single cable with independent control of each waveband for separate analysis or for fusing together on the host processor.

Another important consideration is the spatial resolution of the camera device. There are a variety of multispectral techniques used in cameras. Some use pixel-level filter arrays or multiple optical paths that sacrifice spatial details for spectral diversity.

Pharmaceutical inspection systems demand high spatial resolution per channel to ensure that small defects such as cracks or foreign particles on pill surfaces, air bubbles in liquids, dosage instructions on extrinsic packaging, etc. are clearly identifiable.

Accurate alignment and overlap of the individual spectral bands assist in precisely identifying the position and size of the defects. It also helps to simultaneously trace and correlate the defect characteristics seen through different spectral bands. A multispectral camera with full sensor resolution and a single optical axis for all spectral bands is often the most precise method to achieve such results.

Lastly, builders of future pharmaceutical inspection systems will benefit from new customization technology that allows them to precisely specify the size and location of the spectral bands needed for their particular application. In this way they can keep the number of wavebands to a minimum in order to maximize the efficiency of the system. Having more spectral bands than needed can result in challenging light source requirements and can drastically reduce the speed of the multispectral system.

Vision system builders can use the customization approach to create the right balance between the number of bands, the speed of the system and effectiveness of the inspection process.

FACE MASK INSPECTION MADE BETTER AND FASTER WITH BASLER ACE CAMERAS

CUSTOMER

•  O-Net Industry
•  Location Shenzhen, China
•  Industry: Medical Supply Inspection
•  Implementaion: 2020

APPLICATION

An acute shortage of face masks caused by fear of the spreading coronavirus pandemic has been straining global medical supplies since the start of 2020. The smart face mask inspection system designed by O-Net Industry boosts productivity and increases product conformity rate for the manufacturers. By making the inspection process faster and more effective, this solution can both ease the pressing market need and help face mask manu-facturers drive production cost down.

Headquartered in Shenzhen China, O-Net Industry is a leading company dedicated to machine vision automa-tion. The vision systems designed by O-Net Industry are used in various applications including visual inspections, geometry measurement and OCR among others; they are also able to provide customized solutions tailored to the products to be inspected.

In a traditional production line, a high scrap rate is inevi-table due to interference by environmental factors and the inconsistent working conditions of face mask making machines, resulting in lower efficiency and conformity rate. The application of a vision inspection in the produc-tion process, however, can significantly improve the situation.

All parts of a face mask need to be inspected, including the covering, the edges, the ear loops and the metal strip that lets the wearer bend the mask around the bridge of the nose (Figure 1). Quality control needs to identify and remove masks that are overlapping, broken, contami-nated, askew or in the wrong size.

Face mask inspection is also made more complex by factors including:

•  Uneven illumination occurs during inspection due to the grainy surface of the non-woven fabric of face masks
•  Face masks to be inspected are mmoving and their positions are random on the conveyor
•  The edge, ear loop and metal strip are difficult to distinguish in inspection images

SOLUTION AND BENEFITS

With the help of customized lighting and the Basler ace 5 MP camera, the smart face mask inspection system deve-loped by O-Net can obtain excellent images of each mask. The system can then use the alignment algorithm to check whether the face mask meets standards.

In the inspection process, the system finds the center and corners of the covering part of face masks via the image acquired (Figure 2), to identify products that are miss-hapen. Exact measurement of face masks can also be done. With the center confirmed, the software defines the region of interest (ROI) as well as the baseline, to measure the specific size of a face mask and determine whether it meets standards.

Inspection of ear loops focuses on whether the length of loops and the positions of the fixation points meet the set standards. In the image analysis process, ear loops can be defined as curved lines. The software will detect and extract these curved lines and determine whether they are broken (Figure 6), and if not, calculate their length (Figure 7). The system can detect the fixing points of the ear loops in the image (Figure 8) and measure the dis-tances between the fixation points and their respective neighboring edges, to determine whether they meet standards

Non-woven fabric allows some light to get through, but extra layers can significantly increase its opacity. Thus the folded section of a face mask will appear much darker than the rest in the image. In Figure 4, the upper and bottom part of the face mask appear pale; O-Net’s soft-ware is configured to accept an image where the paler area is 374550 pixels in size. By contrast, the paler area drops to only 28894 pixels, which is almost ten times less, when two face masks overlap (Figure 5). By using such features, O-Net’s system determines whether the face masks are overlapping.

Lastly, the edges of a face mask also need inspection. The system needs to check whether the pitting on the edges is well aligned. Two green baselines are defined based on the outer margins of a face mask. Then the vertical dis-tance from each pitting line to the baselines is measured, so that the system can tell if the pitting on the edges is well aligned

Inspection of the length and position of the metal strip in a face mask is also required. By using the cutomized ligh-ting, the inspection image can show both the metal strip inside and the non-woven fabric wrapping it. Once the ends of the metal strip are found in the image, the length can be calculated (Figure 10). Meanwhile, two baselines are drawn to check that the position of the metal strip is centered

The vision inspection systems developed by O-Net can effectively automate the tedious quality check process and significantly improve product conformity rate. On average, each system can replace up to four skilled human inspectors. In factory applications, the visual inspection system usually runs uninterrupted for long periods; there-fore system stability is essential. O-Net decided on the Basler ace acA2440-20gm camera, due to the well-known stability of this key vision component. Mr. Wang, sales manager of O-Net, explains that “the stability of Basler cameras has helped save considerable mainte-nance costs. Our system development is quite smooth thanks to the Basler pylon Camera Software Suite, as it’s genuinely a developer-friendly software suite, and a short time-to-mark gives us competitive advantages. The vision market is booming in China and speed is vital. Our custo-mers are demanding ever-faster delivery, so the fast and reliable lead time ensured by Basler China is another attractive reason for us to work together.”

The smart software system offers high compatibility and can be customized, as O-Net develops everything from operator interface to architecture. This type of vision inspection software solution can apply to many applications.

TECHNOLOGIES USED

•  Camera: Basler ace acA2440-20gm
•  Lightinh: Customized BT series lighting
•  Software: SV Smart Vision System by O-Net

ALSO READ: THERMAL IMAGING FOR DETECTING ELEVATED BODY TEMPERATURE

TO KNOW MORE ABOUT BASLER ACE CAMERAS FOR FACE MASK INSPECTION INDIA CONTACT MENZEL VISION AND ROBOTICS PVT LTD CONTACT US AT (+ 91) 22 67993158 OR EMAIL US AT INFO@MVRPL.COM

THERMAL IMAGING FOR DETECTING ELEVATED BODY TEMPERATURE

Can thermal cameras be used to detect a virus or an infection? The quick answer to this question is no, but thermal imaging cameras can be used to detect Elevated Body Temperature. FLIR thermal cameras have a long history of being used in public spaces—such as airports, train terminals, businesses, factories, and concerts—as an effective tool to measure skin surface temperature and identify individuals with Elevated Body Temperature (EBT).

In light of the global outbreak of the coronavirus (COVID-19), which is now officially a pandemic, society is deeply concerned about the spread of infection and seeking tools to help slow and ultimately stop the spread of the virus. Although no thermal cameras can detect or diagnose the coronavirus, FLIR cameras can be used as an adjunct to other body temperature screening tools for detecting elevated skin temperature in high-traffic public places through quick individual screening.

If the temperature of the skin in key areas (especially the corner of the eye and forehead) is above average temperature, then the individual may be selected for additional screening. Identifying individuals with EBT, who should then be further screened with virus-specific diagnostic tests, can help reduce or dramatically slow the spread of viruses and infections.

Using thermal cameras, officials can be more discrete, efficient, and effective in identifying individuals that need further screening with virus-specific tests. A variety of institutions, including transportation agencies, businesses, factories, and first responders are using thermal screening as an EBT detection method and as part of employee health and screening (EH&S).

Airports in particular are actively employing FLIR thermal cameras as part of their screening measures for passengers and flight crews. The screening procedures implemented at airports and in other public places are just the first step when it comes to detecting a possible infection: it’s a quick way to screen for anyone who might be sick, and must always be followed up with further screening before authorities decide to quarantine a person.

WHAT FLIR CAMERAS ARE USED FOR THERMAL SCREENING?

While governments outside the United States may choose from many different cameras, FLIR has a 510(k) filing (K033967) with the US Food and Drug Administration (FDA) for select camera models for use as an adjunct to other body temperature screening tools to detect differences in skin surface temperatures. These cameras include the FLIR Exx-Series, FLIR T-Series, FLIR A320, and Extech IR200.

ALSO READ: WHATS THE DIFFERENCE BETWEEN VISION SENSORS AND VISION SYSTEMS?

TO KNOW MORE ABOUT FLIR THERMAL BODY TEMPERATURE SCREENING CAMERAS DEALER MUMBAI CONTACT MENZEL VISION AND ROBOTICS PVT LTD CONTACT US AT (+ 91) 22 67993158 OR EMAIL US AT INFO@MVRPL.COM