Author: Mihir Mehta

Image processing is the use of algorithms and mathematical models to process and analyze digital images. The goal of image processing is to enhance the quality of images, extract meaningful information from them, and automate image-based tasks. Image processing is important in many areas, like computer vision, medical imaging, and multimedia. This article discusses important areas in image processing and mentions Zebra Technologies’ Aurora Design Assistant software.

Understanding Deep Learning (DL) and Its Functioning:

The fields of image processing and Deep Learning (DL) are complementary, especially in the context of computer vision and machine learning tasks. DL is a subset of machine learning, which is a subset of artificial intelligence (AI). DL algorithms are designed to reach similar conclusions as humans would by constantly analyzing data with a given logical structure. To achieve this, DL uses a multi-layered structure of algorithms called neural networks.

The design of the neural network relies on the structure of the human brain. Just as we use our brains to identify patterns and classify different types of information, we can teach neural networks to perform the same tasks on data. DL has succeeded in AI applications, advancing technology and contributing to breakthroughs in computer vision, language understanding, and reinforcement learning.

Widely used deep learning applications:

DL has applications in a vast array of fields, including:

• Image recognition and speech recognition: DL is excellent at image classification, object detection, and facial recognition. It is used for tagging images, recognizing faces for security, and converting speech to text.

• Healthcare: DL is used for medical image analysis, disease diagnosis, and prognosis prediction. It aids in identifying patterns in medical images, such as detecting tumors in radiology scans.

• Autonomous vehicles: DL plays a key role in developing self-driving cars. It uses live data from sensors, cameras, and other sources to decide on steering, braking, and acceleration.

• Manufacturing and industry: DL is applied to predictive maintenance, quality control, and process optimization in manufacturing. It detects defects in products and predicts equipment failures using vision computers.

• Robotics: DL enables robots to perceive and respond to their environments, helping them perform complex tasks.

Deep Learning applications in computer vision:

Computer vision is a part of AI that helps computers analyze and process digital images. It uses algorithms and techniques to make decisions or suggestions based on the images. DL has made significant contributions to computer vision, including:

• Image classification: Categorization of images into predefined classes, fundamental to applications such as object recognition.

• Object detection: Detection of objects within images by providing bounding boxes around them, crucial where it’s necessary to identify and locate multiple objects in a single image, such as in autonomous vehicles or surveillance systems.

• Facial recognition: Key to identity verification, access control, and security. We can accurately identify and match facial features against a database of known faces.

• Image segmentation: Segmentation of images into meaningful regions or objects, valuable in medical imaging for identifying and isolating specific structures within images.

• Medical image analysis: Used in medical imaging tasks, such as detecting and diagnosing diseases from X-rays, MRIs, and CT scans.

• Augmented reality (AR): Enhances the capabilities of AR applications by enabling real-time recognition and tracking of objects.

What is the role of Deep Learning in machine vision?

7 elements of a machine vision system

DL plays a crucial role in machine vision by providing advanced techniques for processing and understanding visual information. Key roles in machine vision include:

• Feature learning: DLs excels at automatically learning hierarchical features from raw visual data, essential in machine vision applications where identifying relevant patterns and features in images is crucial for decision-making.

• Object recognition and classification: DL enables accurate and efficient object recognition and classification. Machine vision systems can use deep neural networks to categorize objects in images, valuable in quality control in manufacturing.

• Object detection: DL is used for object detection tasks in machine vision. It can identify and locate multiple objects within an image, important in robotics and autonomous vehicles.

• Image segmentation: DL techniques are used for image segmentation in machine vision. This involves dividing an image into meaningful segments, useful in medical image analyzing and scene understanding.

• Anomaly detection: DL models can recognize normal patterns and detect anomalies in visual data. Quality control, surveillance, and monitoring systems apply it to identify deviations.

• 3D vision: DL supports 3D vision tasks by processing multiple images or using depth-sensing technologies. Vital in applications like robotic navigation.

• Document and text recognition: DL models are used for optical character recognition (OCR) and document analysis. Aids in automatically extracting information from textual content in images.

• Biometric recognition: DL enhances biometric recognition systems by providing accurate algorithms for face recognition, fingerprint recognition, and other biometric modalities.

How can machine learning benefit image recognition?

Machine learning brings efficiency, accuracy, and adaptability to image recognition tasks, making it a powerful tool for a wide range of applications, such as:

• Automated feature extraction: Machine learning, especially DL, automates the feature extraction process by learning relevant features directly from the data.

• Improved accuracy: Machine learning algorithms perform exceptionally well in image recognition tasks. They can learn hierarchical features, allowing them to recognize patterns and objects in images with great accuracy.

• Adaptability to varied data: Machine learning models generalize well to new and diverse datasets. This adaptability is crucial in image recognition situations where appearances may vary as a result of lighting conditions, angles, and background variations.

• Object detection and localization: Machine learning algorithms enable the classification of objects and the localization of their positions within an image. This is essential for autonomous vehicles, robotics, and surveillance.

• Semantic segmentation: Machine learning techniques can perform semantic segmentation by classifying each pixel in an image. This promotes understanding of the spatial relationships and boundaries between different objects.

Which is the top-rated software for machine vision?

Integrys considers Aurora Design Assistant (Aurora DA) the best software for digital image processing on the market today for the following reasons:

Flowchart-based development: Aurora DA helps you create apps quickly without coding by using flowchart steps for building and configuring applications. Aurora DA offers no-code computer vision —allowing anyone to apply artificial intelligence without having to write a line of computer code. The IDE also lets users design a custom web-based operator interface.

Flexible deployment options: Select your platform from a hardware-neutral environment that is compatible with both branded and third-party smart cameras, vision controllers, and PCs. It supports CoaXPress, GigE Vision, or USB3 Vision camera interfaces.

Streamlined communication: Easily share actions and results with other machines using I/Os and various communication protocols in real time.

Increased productivity and reduced development costs: Vision Academy offers online and on-site training for users to enhance their software skills on specific topics.

Contact Us:

To learn more about Integrys computer vision projects, and products such as Aurora Design Assistant software, or to request a quote, click here to contact us.

With more than 30,000 distinct parts from hundreds of suppliers, a typical new car presents one of modern manufacturing’s biggest challenges. The rapid adoption of innovative new technologies and components like electric drivetrains and sophisticated driver assistance systems isn’t making the manufacturer’s job any easier.

The demand for new solutions has virtually every carmaker and every player in the $2.1 billion automotive parts industry searching for solutions that can help them keep pace with the demand for greater efficiency, higher quality and better traceability. In many applications, the answer is automation.

Automation is hardly new to the auto industry—after all, the industry is widely considered to be the birthplace of the modern assembly line. What’s changing, however, is the penetration of automation technologies into more manufacturing processes. One of the most impactful changes is the widespread deployment of advanced vision systems, using fixed cameras and machine vision systems to streamline data capture and handle sophisticated visual inspections throughout automotive manufacturing.

From a brake component manufacturer’s need to track parts through the supply chain to the electronics manufacturer’s need to perform detailed quality-control inspections to the carmaker’s need for complex 3D analysis, the range of solutions required throughout the automotive manufacturing industry demands a diverse portfolio of hardware and software solutions.

Fortunately, Zebra’s acquisition of Matrox Imaging has created a single-supplier solution with a full range of hardware and software tools to cover almost any vision application. Zebra has partnered with Integrys Limited in Canada to revolutionize the industries with its Machine Vision applications.

Here’s a brief look at some of the end-to-end inspection tasks that automotive manufacturers can accomplish with Zebra’s end-to-end portfolio of machine vision systems:

• Wire Harness Inspection: Today’s passenger cars and light trucks have a dozen or more wiring harnesses, hundreds of connectors, and 2.5 miles of wiring. Leading manufacturers are using machine vision tools to inspect and confirm every wire’s color, gauge, and sequence.
• Pin Inspection: Since the slightest inaccuracies in pin height or alignment can lead to glitchy performance or failure of electronic systems, manufacturers use machine vision solutions to verify that each connector is manufactured to precise specifications before components go on to final assembly.
• Conformal Coating: Innovative machine vision tools can instantly detect inconsistencies like cracks, bubbles, insufficient coverage, incomplete adhesion, and other potential problems in conformal coatings that protect printed circuit boards (PCBs) from corrosion and moisture.
• PCB Inspection: It takes hundreds of PCBs incorporating thousands of microchips and other electronic components to support a modern vehicle. Machine vision technology provides a high-speed, high-precision solution to ensure each critical PCB meets exacting specifications.
• Bead Inspection: Today, machine vision systems evaluate coverage, location, and continuity of the adhesive gaskets on high-speed production lines, detecting many flaws that would escape even the most experienced human inspectors.
• Display Inspections: The number and complexity of electronic displays increase with every new generation of passenger vehicles and light trucks. Machine vision tools can automatically inspect everything from orientation (is it properly installed) and function (is the display properly sequenced) to quality (are there failed pixels) and performance (does it meet standards for brightness, color, and more).
• Color Inspections: Machine vision tools can perform high-speed color inspections to confirm the correct color of everything from body panels and accessories to the color of packaging used for OEM parts that will be shipped to dealers’ service departments.

That’s a diverse list of machine vision applications. Still, it’s only a fraction of what manufacturers can accomplish with Zebra’s impressive portfolio of fixed industrial scanners, machine vision smart cameras, and software tools.

To learn more about the ways Zebra’s Machine Vision solutions, please contact our representative in Canada. Integrys’ advanced Machine Vision systems are reshaping quality control, production efficiency, and automation. Our solutions encompass object location, defect detection, and much more. With 20+ years of experience, we enhance productivity, assure quality, and reduce costs. To learn more about these cutting-edge Machine Vision solutions please contact us by clicking the button below.

Contact Us

Introduction:

In today’s fast-paced technology industry, companies are constantly seeking ways to optimize their systems and improve efficiency. Integrys, a leading engineering firm, recently embarked on a challenging project to help a client enhance the performance of their existing servers without the need for a complete computing system overhaul. With their expertise in needs assessment, design solutions, custom configuration, test and validation, and post-application support, Integrys proved once again why they are at the forefront of innovation.

The challenges in custom design and engineering:

The client’s servers, when operating at full capacity, generated a significant amount of heat due to two main components: two powerful GPUs and an accelerated Network Interface Card from Intel. The excessive heat threatened the stability and longevity of the electronics within the chassis, necessitating a creative solution. Integrys faced several critical challenges in this project:

1. Preserving the Existing Architecture: The project required finding a solution without redesigning the already-built chassis and motherboard and maintaining the positioning of the GPUs and Network Card.
2. Cooling Multiple Heat Sources: Integrys needed to devise a method to bring in cool air from the environment and direct it to three different locations within the system to effectively cool the GPUs and Network Card.
3. Precise Airflow Requirements: The engineering team had to ensure specific flow rates and temperatures were achieved across the pre-existing architecture, which posed further complexities.
4. Seamless Integration: The most demanding mechanical challenge was developing a duct that could be easily inserted into the pre-built chassis without the need for dismantling the entire system.

The Custom Solution:

Led by Electro-Mechanical Designer, Daniyal Jafri, and under the guidance of Engineering Manager, Eric Buckley, the engineering team at Integrys combined their expertise in fluid mechanics, thermal science, and mechanical design to tackle these challenges head-on. By applying scientific principles and innovative thinking, they devised an ingenious solution that achieved the desired flow rates and temperatures while seamlessly integrating into the existing infrastructure.

The team’s initial design iterations involved a compact 80mm duct, which was 3D printed due to the complexity of its geometry. While it met the flow rate requirements, it exceeded the acceptable acoustic noise levels. To address this issue, the team explored a larger 120mm fan, which provided a 4x increase in airflow at the same RPM. Running the 120mm fan at a lower speed to provide that same airflow as the 80mm solution reduced noise levels while still meeting the necessary cooling requirements.

Overcoming the size constraints imposed by the larger fan, the final duct design from Integrys successfully met all the client’s requirements. The innovative design allowed the duct to be seamlessly slid into the system, eliminating the need to remove metal panels within the chassis for assembly. This breakthrough significantly reduced assembly time, saving the client substantial labour costs.

Conclusion:

Integrys has once again showcased its ability to connect technology and innovation. By leveraging their engineering expertise, the team at Integrys devised a ground-breaking solution to enhance the efficiency of the client’s servers. Overcoming the challenges of preserving existing architecture, cooling multiple heat sources, achieving precise airflow requirements, and seamless integration, Integrys delivered a state-of-the-art 3U server solution that exceeded expectations.

Integrys’s dedication and expertise enabled the client to optimize their system without requiring a complete redesign, saving both time and resources. This successful collaboration exemplifies Integrys’ commitment to delivering cutting-edge engineering solutions and reaffirms its position as an industry leader.

Click here to view more engineering services from Integrys or Contact Us for your personalized engineering solution.

The long-standing goal at North Atlantic Industries (NAI) is to accelerate your time-to-mission—to get you to market faster. NAI’s COSA technology, also known as “Configurable Open Systems Architecture” technology, helps you do just that. In a distributed, intelligent, software-driven architecture that allows you to rethink the way you engineer power-critical and I/O-intensive mission systems, COSA satisfies an impressive range of complex and time-critical requirements.

How NAI’s COSA technology works:

• Select I/O boards, single board computers, power supplies or rugged systems to meet your requirements.
• Customize it in modular fashion, selecting from more than 100 available, high-density, fully tested I/O, communications, measurement and simulation smart function modules.
• Leverage NAI’s free software libraries, source-code and comprehensive API to jump-start development and speed your time to test.
• Easily adjust board configuration to add or swap functional capabilities if requirements change.

NAI’s COSA technology works by providing a framework for building software systems that are composed of independent software components, or “modules.” These modules can be configured and combined in various ways to create a customized software solution that meets specific requirements.

The COSA framework provides a set of standard interfaces and protocols that allow the modules to communicate and interact with each other, regardless of the programming language or platform on which they are implemented. This allows for greater flexibility in choosing and integrating different software components, and enables the creation of highly configurable and adaptable software systems.

COSA technology applications include industrial control systems, military systems, and telecommunications networks. It is designed to be highly modular, scalable, and adaptable to changing requirements, making it an attractive solution for complex and dynamic software systems.

COSA modules, boards and systems

Over 100 high-density, COSA smart modules are available for placement on NAI’s intelligent I/O boards, which can operate as standalone systems. Further, NAI’s OpenVPX, VME, cPCI and PCIe boards can be placed into rugged COSA systems that range from a single module to high-density systems supporting up to 10 motherboards and 60 smart modules (and virtually everything between).
 

COSA software and processing advantages

Dedicated FPGAs embedded on NAI’s smart modules provide unique software and processing advantages that drive time and cost out of design, development and qualification schedules—to accelerate your time-to-mission. The programmability, intelligence and self-monitoring capabilities put more I/O capability into the modules themselves so you can reduce the processing load on the SBC and deliver more capabilities at the edges of your applications.

Configurability of NAI’s COTS systems 

• 100+ smart function modules
• Maximum of 6 slots per card
• Maximum of 18 functions per interface unit

NAI’s COSA architecture is massively configurable, providing more than 1.4 quadrillion possible system configurations. That staggering number represents the global population . . . x 2 million.

NAI’s COSA architectures is the most modular, agile, and rugged portfolio of its kind. It enables you to leverage their portfolio of pre-integrated modules, boards, systems and power supplies to quickly and easily meet complex mission processing requirements – today and down the road, allowing you to out-pace, out-adapt and out-last your competition.

Find a team of specialists from Integrys Limited at booth #526 at CANSEC 2023 in Ottawa, Ontario on May 31st and June 1st 2023 to learn more about COSA.

Contact

To learn more about how NAI’s COSA technology can accelerate your time-to-mission or to request a quote, click here to contact us.