Advantages of machine vision applications

Scientifically speaking, the human eye can perceive electromagnetic wavelengths ranging from 390 to 770nm. Machine vision is a technology that converts images into digital signals for analysis and processing, and gives machines visual functions that are beyond the reach of the human eye. It is very suitable for dangerous tasks that are not suitable for human work or occasions where the human eye cannot meet the requirements.

What is Machine Vision?

One of the simplest ways to understand a machine vision system is to think of it as the eyes of a machine. From a professional perspective, machine vision is a technology that uses image processing to achieve automatic detection and analysis applications. It can be said that machine vision is a technical capability that integrates with existing technologies in new ways and applies them to solve real-world problems.

Machine vision is a systems engineering discipline and is sometimes compared to computer vision. In fact, computer vision and machine vision are overlapping technologies. The most significant difference between the two is that machine vision systems require computers and specific hardware and software to work, while computer vision does not require tangible hardware, such as vision boxes or cameras attached to robots. Machine vision can be considered a subcategory of computer vision, with computer vision being its brain. Machine vision cannot work without computer vision.

Specifically, computer vision can analyze images or videos online, as well as images from motion detectors, infrared sensors, or other sources. With the development of edge AI, computer vision has begun to move from the cloud to the edge, closer to the sensors that collect data. Machine vision systems began in the 1950s. From 1980 to 1990, this technology really took off and became increasingly popular.

It is worth noting that with the development of computer vision technology, the possibilities of potential applications of machine vision have also increased accordingly, and have expanded from the main application area - industrial automation environment - to industries such as security, autonomous vehicles, food production, packaging and logistics, and even robots and drones. Now, machine vision systems can inspect and classify a variety of objects and items in various industries, including automotive, electronics and semiconductors, food and beverages, road and vehicle traffic or intelligent transportation systems (ITS), medical imaging, packaging, labeling and printing, pharmaceuticals, etc.

Markets and Markets reports that the machine vision market size is expected to grow from USD 10.7 billion in 2020 to USD 14.7 billion by 2025, at a CAGR of 6.5%. The growing demand for quality inspection and automation, the increasing demand for machine vision systems for non-traditional and emerging applications, and the growing demand for vision-guided robots are the key factors driving the growth of the machine vision market.

Which industries benefit from this?

The advantages of machine vision applications are mainly reflected in five aspects:

One is higher performance and quality in inspection, measurement, metrology and assembly verification.

Second, it can improve the productivity of repetitive tasks, effectively reduce machine downtime and shorten installation time. Third, it has greater flexibility in measurement and metrology, and can ensure stricter process control. Fourth, it can reduce production costs, detect defects early, and reduce scrap rates. Fifth, it occupies a small area, reducing production costs.

At present, the proportion of machine vision applications in the industrial field is very large, and this field has also benefited a lot from it. Through in-depth integration with technologies such as deep learning and machine learning, machine vision can help companies that use this technology better understand data and optimize their business to achieve higher efficiency. For example, BMW uses this technology in combination with artificial intelligence and machine learning to improve efficiency. With the continuous improvement of various technologies, the application areas of machine vision are also constantly expanding, including object recognition, product inspection, appearance size, and even 3D modeling inventory counting. In practice, machine vision systems can also be designed and implemented into the system in a customized way to meet more application requirements.

Based on end use, the machine vision market is segmented into automotive, pharmaceuticals and chemicals, electronics and semiconductors, pulp and paper, printing and labeling, food and beverages (packaging and bottling), glass and metal, postal and logistics, etc. Currently, the automotive industry is an important adopter of machine vision systems. In 2020, the automotive industry accounted for 19.38% and is expected to grow significantly from 2021 to 2028.

In the automotive industry, machine vision is widely used for inspection purposes, including presence inspection, error proofing, assembly verification, and final inspection. In addition, machine vision systems are used for dimensional measurement, robot guidance, and test automation, which are measurement and guidance applications. As a result, the demand for mechanized imaging is high across the automotive industry and is expected to continue to grow steadily in the coming years.

The “eyes” of machine vision

All machine vision methods are inspired by the human visual system to extract conceptual information from two-dimensional images - they have 2D image-based capture systems and computer vision algorithms that mimic human visual perception. Humans perceive the world around them in 3D. Among the three classifications of machine vision systems, 1D vision systems do not look at the entire picture at once, but analyze signals one line at a time. They usually detect and classify defects in products manufactured in a continuous process, such as metal, plastic, paper, non-woven sheets or roll products.

Machine vision systems typically use regular 2D imaging under standard lighting conditions. Sometimes objects require specific lighting to record defects - for example, machine vision systems can use multispectral imaging, hyperspectral imaging, infrared bands, line scan imaging, 3D imaging, and X-ray imaging. Typical 2D visible light illumination images are monochromatic compared to more complex lighting, which takes into account factors such as color, frame rate, resolution, and whether the imaging process is synchronized across the image, making it suitable for systems that need the technology to track specific moving items.

Currently, there is no typical machine vision system that can serve as a reference for other designs, because machine vision is a capability rather than a product or a specific type of design. In actual applications, they are implemented by integrating different components together. The main components of a machine vision system include lighting systems, lenses, image sensors, visual processing, and communication systems. Lights illuminate the parts to be inspected, making their features stand out so that the camera can see them. The lens captures the image and presents it to the sensor in the form of light.

The sensor converts this light into a digital image and sends it to the video processor for analysis. Vision processing includes algorithms that inspect images and extract the information needed for necessary inspection and decision-making. If machine vision gives the machine an extra pair of eyes, the image sensor is the "vision" of the machine vision system in a literal and practical sense, which is equivalent to the "eyes" of the system. Its quality is directly related to the level of "vision" of the entire system. With the increasing application of machine vision, image sensors have also ushered in a good opportunity for development. According to data provided by ON Semiconductor, the machine vision sensor market will grow at a CAGR of 14% from 2018 to 2022.

CMOS image sensors are the most common type of sensor used in machine vision. Like CCD sensors, they can be either monochrome or color. The ON Semiconductor MT9P031I12STC-DR1 is a color CMOS sensor for high-resolution machine vision applications. It belongs to the MT9P031 family of CMOS digital image sensors. This product family features an active imaging pixel array of 2592 H x 1944 V. It is a programmable, highly integrated product with low power consumption. It has the image quality of a CCD while maintaining the size and cost advantages of a CMOS image sensor. The MT9P031 sensor can operate in default mode or be user-programmed for frame size, exposure, gain settings, and other parameters. The default mode outputs full-resolution images at 14 frames per second (fps).

Of course, some machine vision applications also have specific resolution requirements. Unlike traditional RGB viewing applications driven by display standards (16:9 or 4:3), many robotics and machine vision solutions can be optimized by using different resolutions. For example, ON Semiconductor's 2-megapixel AR0234 provides additional pixels in the X direction for a better barcode solution. In the XGS series of image sensors, 8-megapixel, 9-megapixel, and 12-megapixel products provide image ratios of 2:1, 1:1, and 4:3, respectively. XGS's 30-megapixel and 320-megapixel products provide image ratios of 1:1 and 4:3, respectively - 1:1 can be used in semiconductor inspection, while 4:3 can be used in screen inspection.

3D Machine Vision Solutions

3D machine vision image detection is closer to human eyes. With the help of digital 3D scanning data, the size of an object can be extracted, including surface area, volume and shape size. 3D vision sensing technology is a depth sensing technology that enhances the camera's ability to perform facial and target recognition. There are currently three mainstream 3D optical vision solutions on the market, namely: Stereo Vision, Structured Light, and Time of Flight (ToF).