To answer this question, we can break down at a machine vision system to analyze one by one.
A simple machine vision systems include
- Visual imaging portion
- The image processing section
- Motion control section
1. vision imaging portion
And several typical vision imaging assembly comprising: a light source, camera, industrial camera.
The light source and the optical lens requires us to master knowledge, different ways of lighting, allowing the camera to produce a completely different image of the object; and selecting magnification lens focal length, such as different vision will determine the fidelity imaging. For a machine vision engineer, learn how to choose the camera, how to select the source, how to determine the way lighting is the most basic skills.
Industrial cameras we need to grasp the photoelectric knowledge, grasp the difference between a camera sensor, master the basic knowledge of imaging such as image resolution, dynamic range, field of view and so on, so that we can choose the right camera according to the needs and scenarios, most quickly grasp these the method is to study the relationship of knowledge through these imaging parameters and imaging buy entry-level SLR
The image processing section
We generally understand the image processing is performed on the PC machine, in fact, in the industrial field, most of the IPC, because it is stable, coupled with cost advantage.
In recent years the development of embedded hardware is also booming, many factories for the needs of small states such as control switches and hundreds dashboard monitor, can use Raspberry Pi and other open source hardware.
For starters, you can master the development priority PC platform, X86 platform, after the familiar can be extended to embedded platforms.
In the software part, most of the application layer using C # ,. net, QT, C ++ to achieve, so grasp of which a programming language is mandatory; in the image algorithm level, the typical open-source algorithm opencv, there are commercial halcon, visionpro, it is recommended beginning can start to halcon entry; if you want further, you can look at machine learning, this may be the main direction of the future in the algorithm level.
For the theory, more of a grasp the basic concepts of image processing, the book "machine vision algorithms and applications", worth a visit.
3. The motion control section
Typical motion control card such as solid high, you can start to look at. A little more advanced PLC, can also play up this part of the difficulty is that the precision of the correction, because a lot of scenes and demand accuracy requirements are very high.
In addition to the above three points, to build the capacity of the overall program is crucial because programs need to string together these parts, and real-life scenarios and can be linked to actual production to meet the automation needs.
The overall program will depend on the ability to build
- Depth understanding of the production process
- We have a deep understanding of the interface between all components of relationship
Author: Hao Ge world
Source: Jane book
Jane book copyright reserved by the authors, are reproduced in any form, please contact the author to obtain authorization and indicate the source.