1. Visual Imaging Section
Visual imaging comprises several typical components: light source, lens, and industrial camera.
Both light sources and lenses require a grasp of optics. Different lighting techniques can produce completely different images of objects; and the choice of lens magnification, focal length, and field of view directly determines the realism of the image. For a machine vision engineer, mastering how to choose lenses, how to select light sources, and how to determine lighting techniques are the most fundamental skills.
Industrial cameras require us to have knowledge of optoelectronics, understand the differences between camera sensors, and grasp the basics of image imaging such as sharpness, dynamic range, and field of view. Only then can we choose the right camera based on our needs and the scenario. The fastest way to master this knowledge is to buy an entry-level DSLR and thoroughly study the relationship between these imaging parameters and imaging.
2. Image Processing Section
Image processing is generally understood to be done on PCs, but in the industrial field, industrial control computers are mostly used because they are stable and have cost advantages.
In recent years, embedded hardware has also been developing rapidly. Many factories can fully utilize open-source hardware such as Raspberry Pi to meet small needs, such as controlling the switching and status monitoring of hundreds of dashboards.
For beginners, it is advisable to first master the development of the PC platform and the x86 platform, and then extend to the embedded platform after becoming familiar with them.
In the software part, most application layers are implemented using C#, .NET, QT, and C++, so mastering one of these programming languages is essential. In terms of image algorithms, typical open-source algorithms include OpenCV, while commercial ones include Halcon and VisionPro. It is recommended to start with Halcon. If you want to go deeper into the algorithm level, you can study machine learning, which may be the main direction in the future.
In terms of theory, it's more about mastering the basic concepts of image processing.
3. Motion control section
Typical motion control cards like Googol are worth exploring. More advanced PLCs can also be used; the challenge here lies in precision calibration, as many scenarios and requirements demand extremely high accuracy.
In addition to the three points mentioned above, the ability to build an overall solution is crucial, because the solution needs to connect all these parts and be able to connect with real-world scenarios to meet actual production automation needs.
The ability to build a complete solution depends on a deep understanding of the production process and a profound understanding of the connections and relationships between all components; both of these require experience accumulated from multiple projects in order to provide a good solution.
