Machine vision is to use machines instead of human eyes do measure and judge. Sent to a dedicated image processing software, the information distribution of pixels and the brightness, color, etc., into a digital signal; image processing software for the signals various operations to extract features of the target, and then to control the device operation field in accordance with the results of discrimination . With the vigorous development of China's manufacturing industry, China is to become one of the most active international machine vision development of the region. Large-scale use of the category from the beginning of the electronics, pharmaceutical and other occupations, and gradually extended to the packaging, cars and other areas, and there are a lot of use in transportation, printing and other professional. Simply use the scale include various occupations of the national economy. How to choose a suitable machine vision software has become the automation Health to do a loop, the majority of enterprises due to visual junior, not the system to the user joined the line to explain clearly. In this paper we focus on and select a few key points about the machine vision software, as well as introduce some commonly used vision software.
First, the machine vision software selection points
1, accuracy locator
Or accurate positioning of the target function is an important feature of a system or detected by the visual motion guide system. The traditional choice is the gradation object location value related to identify objects. Although this skill has been widely used, but its image quality deterioration in the situation on the lack of stability. Image quality deterioration may be due to factors affecting the clutter, the brightness is not the same and hide and so on. In contrast, the geometric targeting method is a new way to use it outlines goals to identify and target characteristics. Most dimensional visual image self-developed image processing software using geometric targeting method, and high efficiency, can effectively help users solve the location problem.
2, tool library or application software
Machine vision software mainly to the sale of two typical ways: one is the tool library contains a variety of processing algorithms, and other application software is designed to achieve a particular type of work. Both have advantages and disadvantages, it is necessary to confirm the first time this feature before making a choice. Image processing software provides dimensional visual image that contains both, both can provide professional applications to meet the diverse and changing needs of customers.
3, programming and easy to operate
Concise, intuitive graphical interface is simple to use and set point. The main difference between today's machine vision product lies in their graphical interface. Interface probably be assessed from two aspects "Settings" and "Operation." For an engineer is concerned, it is probably very complex, but it should be on a very simple operator.
4, sub-pixel accuracy
Resolution visual system is able to distinguish the smallest feature of the system. For example, a 'visual scale (FOV) using a computer image of 640x480 pixels of resolution obtained is 1/640 or 0.00156' '. In fact, the machine vision algorithms with sub-pixel talent. In other words, these algorithms can measure or draw smaller than a pixel unit.
5, easy to upgrade
Machine vision systems can be used in various occasions, they can use the size of the camera from fashion to supervision system. When the selection of a system should be thinking of the future system upgrades. Vision software system based on general-purpose very well cut. Based on the end-user about additional camera, lighting changes, such as changes in visual things to think about the future needs of the system.
6, the image pre-processing algorithm
Most image processing software provides pre-processing algorithms, it is necessary to compare their offers richness and efficiency.
7, system integration
If you are not very proficient in machine vision skills, then for your project you will need a system integrator, ideal visual merchandise can withstand a wide range of system integrators.
Second, the commonly used machine vision Software
1. OpenCV, a free open source image processing library
OpenCV open source is introduced in recent years, free computer vision library, using the function it contains can easily achieve digital image and video processing. While taking advantage of the object-oriented programming tools VC ++ 6.0, a program written in C ++ algorithm library OpenCV provides great convenience for digital image processing VC ++ programming, it will become a powerful tool for video processing images.
Cons: Because it is open source software, so its many versions, the library complex, the efficiency of applications received, more suitable for research and learning, is not suitable for industrial applications.
2. Halcon, powerful image processing library
HALCON is the German MVtec developed a comprehensive standard package of machine vision algorithms, with a wide range of machine vision applications integrated development environment. It saves production costs and shorten the software development cycle --HALCON flexible architecture facilitates machine vision, rapid development of medical images and image analysis applications. In Europe and Japan, the industry is already recognized Machine Vision Software having the best performance
Disadvantages: more expensive, every distribution need to re-purchase authorization.
3. NI Vision, fast verification of image processing library (including visual aides, VBAI)
Graphical programming and NI vision software engineering vision algorithms can be implemented in higher development efficiency. To choose a configurable environment, and comprehensive programming library, to better meet the needs and quick start. They may be used in conjunction with any combination of hardware, good compatibility, making it easy to use existing code, manage and maintain a plurality of hardware systems.
Cons: A nice necessarily expensive.
4. Matlab, the underlying algorithm verification choice
matlab convenient usage, particularly integration of a number of image processing functions, virtually every aspect of image processing methods used. The common operations have made the corresponding built-in functions, so that users do not have to consider how to read images, the minutiae of how to convert color space and other fixed and sophisticated algorithms, the study will focus put into the algorithm.
Drawback: matlab to grasp the details of treatment is not enough.
Letter to switch to AI students in school
[AI] complete AI self-learning course, the most detailed resource consolidation!
AI switch need to look at some of the articles
[PDF] to send the book Python programming from entry to practice
Python from entry to the master, the depth of learning and machine learning materials spree!
Press scan code tease returnees