(For more details about the above companies, click to read the original text)
The hot topics in the field of machine vision in 2018 include: 3D imaging, collaborative robots, deep learning (and more broadly, machine learning and artificial intelligence), embedded vision, multispectral imaging, hyperspectral imaging, and polarization imaging.
As the industry’s attention and demand for machine vision usage continue to rise, more production technologies and products are being developed and released in this field. This article will mainly discuss three trending topics in the machine vision field, as well as some of the latest product information released in these areas.
Figure 1: Polarization cameras used to detect smooth and curved surfaces in images, indicating that measurements in these areas may be invalid, including intensity images (right) and false-color polarization (left).
Deep learning is a field of machine learning that enables computers to be trained and learn through structures like convolutional neural networks (CNNs). Of course, deep learning has been a hot topic in recent years. Currently, many companies in the machine vision market have invested a lot of time and resources into developing deep learning products. Recently developed, announced, or upgraded deep learning software products or tools include Cognex (VisionProViDi), CythSystems (NeuralVision), EVT (EyevisionSoftware), LaonPeople (NAVIAI), Mathworks (MATLAB), MatroxImaging (MatroxDesignAssistant), MVTec (HALCON), and SUALAB (suaKIT). Some companies have also developed components related to deep learning, including cameras and image acquisition cards. These companies include but are not limited to CEVA (NeuProAI processor for deep learning), Euresys, FLIR Integrated Imaging Solutions (Firefly deep learning camera), and Silicon Software (deepVCL image acquisition card).
In addition, companies like Amazon, Ambarella, Google, Intel, and Microsoft have recently released or announced products or platforms related to deep learning.
Polarization imaging sensors and polarization cameras can be used to detect hidden material characteristics that traditional imaging techniques cannot detect. While this is not a new technology, polarization cameras and imaging components began to become more mainstream in the market in 2018. This can be at least partially attributed to Sony’s launch of the IMX250MZR/MYR (monochrome, color) CMOS image sensor, a 5-megapixel global shutter CMOS image sensor with a four-way polarization filter design, featuring four separate polarization filters arranged at angles of 0°, 45°, 90°, and 135° on the sensor. Since the launch of the IMX250MZR/MYR polarization sensor by Sony, many companies have announced the introduction of cameras based on this polarization sensor, including Allied Vision, Baumer, Flir Systems, JAI, LUCID Vision Labs, Matrix Vision, Photonfocus, Pixelink, and Teledyne DALSA, with Sony also launching its own polarization camera based on this sensor. Recent manufacturers releasing polarization cameras include Teledyne DALSA (Piranha4 line scan polarization camera), Photron (Crysta2D polarization camera), and 4D Technology Corporation (PolarCam snapshot micro-polarizer camera). By the time this content is published, I expect there will be other companies releasing polarization camera products.
Embedded vision can be said to be the hottest topic this year, in fact, this topic has received a lot of attention over the past few years. With technological advancements, embedded vision systems are gradually reducing in cost, size, and power consumption, and machine vision and image processing technologies are expected to enter thousands of new emerging applications. Many machine vision camera companies have realized this and have either expanded into the embedded vision market or increased their development efforts in the embedded vision market. Here are some companies that have developed embedded vision cameras and recently launched relevant products: Alkeria, Allied Vision, Basler, Baumer, D3 Engineering, e-con Systems, FLIR IIS, IDS Imaging Development Systems, Jadak, LUCID Vision Labs, Matrix Vision, Omron Microscan, OpenMV, The Imaging Source, Sony, XIMEA, and Vision Components. In addition, many companies have also released embedded vision products over the past year, such as PCs, motherboards, development kits, sensors, graphics processing units, controllers, and software, including but not limited to Allied Vision, Ambarella, AAEON, Basler, Cadence, CEVA, Critical Link, EPIX, FRAMOS, Logic Supply, Matrox Imaging, MVTec, Neousys Technology, NVIDIA, Sony, Vecow, and Xilinx.
David Dechow, Chief Vision Systems Architect at Integro Technologies, stated in an interview with Vision Systems Design: “Embedded computers have seen explosive growth. A few years ago, we referred to embedded computers as industrial PCs, which was a very cutting-edge technology at the time; but now we see a trend of integrating machine vision and deep learning in both single-board machines and system solutions.” He also added, “It is truly exciting to be able to integrate everything needed for machine vision and deep learning into an industrial or embedded processor.”
Source: China Vision Network, this article only represents the original author’s views, the reprint is for sharing industry information, with deletions and modifications. If there is any infringement, please contact the association staff for processing, thank you!