What does machine vision mean? In simple terms, it can be said that it is the processing of image information using computer technology. The object of computer vision can be practically anything – the human face, biological samples, X-rays, traffic density, human activity (especially the production process) and more. A characteristic feature of machine vision systems is the connection of the camera system to the production process with an orientation to typical tasks, such as input / output quality control, shape control, object counting, color product control, barcode reading, defect detection, etc.
How do machine vision systems work?
The camera captures the image of the scanned object, the image analysis system evaluates it according to a precisely defined algorithm and performs the required action based on the evaluation result. Today, the only effective and practically possible way to evaluate image information is a computer. It is typical for machine vision that the exchange of information obtained with the environment takes place on the basis of feedback with the production process. On the one hand, we obtain the input information. On the other hand, various actions perform based on the results of image analysis. Such as discarding an unsuitable or poorly manufactured product, accuracy when filling bottles, checking the presence of labels, and more. The information about the fulfillment of the required action goes directly to the device via digital inputs. The information pass to the control system, which already performs the necessary intervention.
Development
A few years ago, machine vision systems had a reputation for complex, expensive and unreliable technologies. With the development and development in the field of image sensors and especially the performance of today’s processors, it brought the possibilities of machine vision closer to the requirements of users. Nowadays, more and more emphasis place on the quality of each piece produced. Failure of the human factor cannot be ruled out, whereas machine vision systems can operate 24 hours a day. In most cases, image analysis systems thus become the best and only possible solution.
The principle of machine vision
Machine vision “approaches” a given task very much like a human being.The human eye captures an image of a subject. And the system evaluates it according to a prescribed algorithm and performs an action based on the evaluation result. For the purposes of the technical description, it is easiest to look at a machine vision system as a general sensor.
The object must be able to reflect this radiation. So the reflected radiation creates a bright two-dimensional image on the sensor’s sensing element. It is important that this image contains the information you need to know about the object monitored. This condition seems to be completely trivial, but it is on which many tasks have already failed. The most important part of the machine vision system is the ability to meet this condition in the real world at a reasonable cost.
Applications
The monitored object irradiates with visible light coming from artificial sources – illuminators. A two-dimensional brightness image is created on the camera’s sensing element (a camera here means a system consisting of a
machine vision lens and the necessary electronic circuits). In a camera with a semiconductor sensing element provided by the image sensor are light-sensitive elements. Measurement quantities can carry information reduced to the combined brightness of the colored components of the incident light (
black and white cameras). They can also carry information about the color of the object under investigation (color cameras).
In fact, the only possible way to evaluate image information that can work in practice today is computer processing. It is necessary to convert analog measuring quantities into digital form – digitize the image. The information in digital form process the computer by suitable algorithms. The obtained information is again in digital form. And it is transmitted to the environment via digital outputs or digital communication interface.
It is typical for machine vision that the exchange of system information with the environment acts as feedback in controlling the
production process. The system obtains the necessary input data from the process. Typically, this is the command to take a picture when the subject is in the appropriate position. Depending on the result of the image evaluation, the system usually performs an action on the process. The intervention can mediate directly the digital outputs of the machine vision system.