Machine vision software is the intelligent and most crucial component of an inspection system. The choice of software determines the time required to write and debug the inspection program, the performance of the inspection operation, and so on.
1. Camera and optical components
This type typically contains one or more cameras and lenses (optical components) for capturing images of the object being inspected. Depending on the application, the camera may be based on the following standards: black and white, composite color (Y/C), RGB color, non-standard black and white (variable scan), progressive scan, or line scan.
2. Lighting
Lights are used to illuminate components so that better images can be captured by the camera. Lighting systems can come in various shapes, sizes, and brightness levels. Common forms of lighting include high-frequency fluorescent lamps, LEDs, incandescent lamps, and quartz-halogen optical fibers.
3. Component Sensors
It typically takes the form of a grating or a sensor. When the sensor detects a part approaching, it sends a trigger signal. When the part is in the correct position, the sensor tells the machine vision system to acquire an image.
4. Image acquisition card
Also known as a video capture card, this component is usually a card that plugs into a PC.
This capture card connects the camera to the PC. It acquires data (analog or digital signals) from the camera and then converts it into information that the PC can process.
It can also provide signals to control camera parameters (such as trigger, exposure time, shutter speed, etc.). Image acquisition cards come in many forms and support different types of cameras and different computer buses.
5. PC platform
Computers are a key component of machine vision .
In image detection applications, Pentium or higher CPUs are typically used. Generally speaking, the faster the computer, the shorter the time it takes for the vision system to process each image.
Because manufacturing sites are often exposed to vibrations, dust, heat radiation, and other factors, industrial-grade computers are generally required.
6. Detection software
Machine vision software is used to create and execute programs, process acquired image data, and make pass/fail decisions.
Machine vision comes in many forms (C language library, ActiveX control, click programming environment, etc.), and can be single-function (e.g., designed only to detect LCD or BGA, alignment tasks, etc.) or multi-function (e.g., designed as a kit that includes measurement, barcode reading, robot navigation, field verification, etc.).
7. Digital I/O and Network Connectivity
Once the system completes this detection part, it must be able to communicate with the outside world, for example, to control the production process or send "pass/fail" information to the database. Typically, a digital I/O card and/or a network interface card (NIC) are used to enable communication between the machine vision system and external systems and the database.
Configuring a PC-based machine vision system requires careful planning and attention to detail to ensure your inspection system meets your application needs. Here are a few points you should consider:
1. Define your goal. This is perhaps the most important step: deciding what you need to achieve in this detection task. Detection tasks are generally divided into the following categories:
Measurement or metrology
Read character or encoded (barcode) information
Detecting the state of an object
Recognizing and identifying special characteristics of pattern recognition
Compare or match objects with templates.
2. The navigation and inspection process for a machine or robot can consist of a single operation or multiple tasks related to the inspection task.
3. To confirm your task, you should first clarify the tests you need to perform to maximize the detection of the components, that is, the defects you can anticipate.
4. To clearly define which tests are most important, it's best to create an evaluation table listing the "must-do" and "can-do" tests. Once the main tests are satisfactory, more tests can be added to improve the testing process. Remember, adding tests will increase testing time.
5. Determine the speed you need – how long does the system need to detect each component?
This is not only determined by the speed of the PC, but also affected by the speed of the production line.
Many machine vision systems incorporate clocks/timers, allowing for accurate measurement of the time required for each step of the inspection operation. This data enables us to modify our programs to meet time constraints. Typically, a PC-based machine vision system can inspect 20-25 parts per second, depending on the number of parts inspected, the processing program, and the computer's speed.
8. Choose your hardware wisely.
The performance of a machine vision system is closely related to its components. There are many shortcuts in the selection process, especially in optical imaging, which can significantly reduce system efficiency. Below are some basic principles you must keep in mind when selecting components.
1. Camera: The choice of camera is directly related to the application requirements, and usually three points are considered:
Black and white or color;
Movement of components/targets;
Image resolution.
In most inspection applications, monochrome cameras are used because monochrome images provide 90% of the visible data and are cheaper than color images.
Color cameras are primarily used in applications requiring the analysis of color images. The choice between a standard interlaced scan camera and a progressive scan camera depends on whether the component moves during inspection. Additionally, the image resolution must be high enough to provide sufficient data for the inspection task. Finally, the camera must be of high quality and unaffected by vibration, dust, and heat found in industrial environments.
2. The crucial factors of optical components and lighting are often overlooked.
When you use poor optics or lighting, even the best machine vision system will perform worse than a low-capacity system with good optics and proper lighting.
The goal of optical components is to produce the best and largest usable area of an image, and to provide the best image resolution. The goal of illumination is to illuminate the key features of the part that needs to be measured or inspected. Typically, the design of an illumination system is determined by factors such as color, texture, size, shape, reflectivity, etc.
3. Although the image acquisition card is only one component of a complete machine vision system, it plays a very important role.
The image acquisition card directly determines the camera's interface: black and white, color, analog, digital, etc.
The goal of using an analog input image acquisition card is to convert the image captured by the camera into digital data as inconsistently as possible. Using an incorrect image acquisition card may result in erroneous data.
Industrial image acquisition cards are typically used for inspection tasks. Multimedia acquisition cards, because they modify image data through automatic gain control, edge enhancement, and color enhancement circuitry, are not used in this field. The goal of image acquisition cards using digital input is to convert and transmit digital image data output from a camera to a PC for processing.
Considering various changes:
Human eyes and brains can recognize targets under diverse conditions, but machine vision systems are not so versatile; they can only function according to their programmed tasks. Understanding what your system can and cannot see can help you avoid failures (such as mistaking a good part for a bad one) or other detection errors. Factors to consider generally include part color, ambient light, focus, part position and orientation, and significant variations in background color.
Choose the right software:
Machine vision software is the intelligent and most crucial component of an inspection system. The choice of software determines the time required to write and debug the inspection program, the performance of the inspection operation, and so on.
Machine vision software offers graphical programming interfaces (often called "Point & Click"), which are generally easier to use than other programming languages (such as Visual C++), but have limitations when you need specific features or functionalities. Code-based software packages, while very difficult and requiring coding experience, offer greater flexibility in writing complex, application-specific detection algorithms. Some machine vision software provides both graphical and code-based programming environments, offering the best of both worlds and providing considerable flexibility to meet diverse application needs.
Communication and recording data:
The overall goal of a machine vision system is to achieve quality inspection by distinguishing between good and bad parts. To achieve this, the system needs to communicate with the production line so that it can perform certain actions when a bad part is detected. These actions are typically performed through digital I/O boards connected to PLCs in the manufacturing line, allowing the bad parts to be separated from the good ones.
As an exception, machine vision systems can connect to networks, allowing data to be transmitted to databases for logging and for quality control personnel to analyze why defective products occur. Careful consideration at this stage will help in seamlessly integrating machine vision systems into production lines.
The issues that need to be considered are:
What type of PLC was used, and what was its interface?
What type of signal is needed?
What type of network are you currently using or must use?
What file format is used for transmission over the network? Typically, an RS-232 port is used to communicate with a database to record data.
Preparing for the future:
When selecting components for a machine vision system, always keep in mind future production needs and potential changes. These will directly impact the ease with which your machine vision hardware and software can be modified to meet new future tasks.
Advance preparation will not only save you time, but also reduce the overall system cost by reusing existing inspection tasks in the future. The performance of a machine vision system is determined by its weakest link (just as the capacity of a barrel is determined by its shortest stave), while its accuracy is determined by the information it can acquire. Investing time and effort in properly configuring the system can build a trouble-free and resilient vision inspection system.
