For robots to be able to perform more complex tasks, they not only need better control systems, but they also need to be more aware of changes in the environment. Among them, the Miracleman vision has become the most important perception function for robots with its large amount of obtainable information and wide range of information.
Machine vision technology is a cross-discipline involving artificial intelligence, neurobiology, psychophysics, computer science, image processing, pattern recognition, and many other fields. Machine vision mainly uses computers to simulate the visual function of human beings, but it is not just a simple extension of the human eye, more importantly, it has a part of the function of the human brain - extracting information from the image of the objective object, processing and understanding, and ultimately used for actual detection, measurement, and control.
The so-called machine vision is defined as the automatic reception and processing of images of a real object using optical devices and non-contact sensors, to obtain the required information or for the control of the robot's movement.
In the 1970s, several practical vision systems appeared for applications such as integrated circuit production, precision electronics, industrial robotic product assembly, and inspection of beverage canning quality. By the late 1980s, specialized image processing hardware appeared, and people began to systematically study the robot vision control system. With the enhancement of computer functions and the price decline, as well as the rapid development of image processing hardware and cameras, the robot vision system has attracted more and more researchers. At present, machine vision is an indispensable technology in the modern processing and manufacturing industry, which is widely used in food and beverage, cosmetic, pharmaceutical, building materials and chemical, metal processing, electronic manufacturing, packaging, automobile manufacturing, and other industries, for example, the visual inspection of printed circuit boards, automatic flaw detection on the surface of steel plates, and the detection of container volume or impurities, etc, which are all applying to machine vision technology. The application of machine vision technology makes the machine work more and more instead of human labor, which undoubtedly improves the level of production automation and the intelligence level of the inspection system to a great extent.
Characteristics of machine vision
（1）High precision. A good machine vision system is capable of spatially measuring one of 100 or more targets with a high degree of accuracy because such measurements do not require contact with the target, are not damaging or dangerous to the target, and utilize computer technology.
（2）Continuity. Machine vision systems can save people from fatigue. Because there is no human operator, it also changes the operation changes due to human reasons.
（3）Flexibility. Machine vision systems are capable of performing a variety of different information acquisition or measurements. When the application requirements change, the software can be adapted to the new requirements by simply changing or upgrading the software accordingly.
（4）Standardized. The core of machine vision systems is visual image technology, so the standard of machine vision system products from different manufacturers is consistent, which provides great convenience for the wide application of machine vision.
Components of the vision system
A machine vision system is a system that captures an image by a machine vision sensor, transmits the image to a full processing unit, and digitally processes the image to discriminate size, shape, color, etc. based on the pixel distribution and information such as brightness and primary color, and then controls the action of on-site equipment based on the results of the discrimination. Taking the automobile vehicle size machine vision measurement system as an example, the machine vision system generally consists of a lighting system, vision sensor, image acquisition card, image processing software, monitor, computer, communication unit, and so on.
（1）Vision Sensors
The vision sensor is the direct source of information for the entire machine vision system, which mainly consists of one or two graphic sensors, sometimes accompanied by light projectors and other auxiliary equipment. The main function of the vision sensor is to acquire enough raw images for the machine vision system to process.
A vision sensor is an instrument that utilizes an optical element and an imaging device to acquire image information about the external environment, and the performance of a vision sensor is usually described in terms of image resolution. The accuracy of a visual sensor is not only related to the resolution but also to the detection distance of the measured object. The farther away the object to be measured is, the worse its absolute positional accuracy is.
A vision sensor has thousands of pixels that capture light from an entire image. The clarity and detail of an image are usually measured in terms of resolution, expressed as the number of pixels. After capturing an image, the vision sensor compares it to a baseline image stored in memory to make an analysis.
（2）Image capture card
Image capture card is an important part of the machine vision system, its main function is the camera output data for real-time acquisition. An image capture card, also known as an image capture card, is a hardware device that can acquire digital video image information, and store and play it out. Many image capture cards can capture video information while obtaining accompanying audio so that the audio portion and the video portion are synchronized for preservation and playback at the time of digitization.
（3）Light source
The core of a machine vision system is image acquisition and processing. All information comes from the image, and the quality of the image itself is extremely critical to the entire vision system. The light source is an important factor affecting the image level of the machine vision system, as it directly affects the quality of the input data and at least 30% of the application effect.
Through appropriate light source illumination design, the target information and background information in the image can be optimally separated, which can greatly reduce the difficulty of image processing algorithms segmentation and recognition, and at the same time, improve the positioning and measurement accuracy of the system, so that the reliability and comprehensive performance of the system can be improved.
There are at least a few roles for a light source in a machine vision system:
1. to illuminate the target and increase its brightness;
2. to create an imaging effect that is most conducive to image processing;
3. to overcome the interference of ambient light to ensure the stability of the image;
4. to be used as a measurement tool or reference
（4）Calculator
The computer is a key component of machine vision. The image information obtained by the vision sensor is stored and processed by the computer, and the processed results are output according to various purposes. The faster the computing speed of the computer, the shorter the time taken by the vision system to process the image. Industrial-grade computers are generally required because of the vibration, dust, and heat radiation that are often present at manufacturing sites. In addition to displaying graphics on a monitor, images can be output using a printer or plotter.
Lenses and vision sensors
The camera is the main component of the vision system, i.e., the optical element lens and the vision sensor.
（1）Lenses
There are two types of optical element lenses, fixed focal length and variable focal length lenses. Fixed focal length lenses are suitable for situations where the position of the target is fixed. In this case, the camera uses the fixed installation method. The advantages of fixed focal length lenses are good imaging quality, small mass, small size, inexpensive, and so on. The disadvantage is that the adjustability is poor, and can not change the field of view range. A variable focal length lens is suitable for the requirements of the field of view range variable camera system, such as a weld tracking system. This is to control the aperture, zoom, and focus. Therefore, the corresponding control circuit should be added.
（2）Vision Sensors
There are many types of vision sensors, such as phototransistors, laser sensors, photoconductive camera tubes, and solid-state camera devices. However, only two of them are applicable in the field of industrial robotics, i.e., photoconductive camera tubes and solid-state camera devices. Photoconductive camera tubes are the first image sensors to be used.
RZ Optics Co., Ltd. is committed to producing professional optical components, from design to production to sales integration, the main products include spherical lenses, including optical domes, plano-convex lenses, optical windows, optical prisms, mirrors, and so on. We also could provide OEM service, and use many kinds of substrates such as sapphire, fused silica, silicon, etc. Our optical domes, optical prisms, mirrors, etc. can be used for the camera and vision sensor applications mentioned in this article. Please contact us if you have any needs!
Next: optical viewfinder
