In the world that we live in today tasks need to be performed quickly, efficiently and consistently, all in an effort to save time and money. Nowhere is this truer than in the manufacturing industry.
Production lines are commonly used to speed up the manufacturing process. The challenge with a production line is to automate it as far as possible, but still ensure that wastage is minimised and quality is maximised.
The most effective way of minimising waste and maximising quality is to implement some type of control system. Traditionally some type of scanning device is used to check various parameters of a product as it moves through the production line. However these scanners are usually quite expensive and very task specific.
The use of scanners in a production line is to some extent becoming outdated and redundant due to the fact that they can only be used to perform specific tasks. The more modern approach to minimise wastage and maximise quality is by the use of a machine visioning system.
What is machine vision?
It is the automatic extraction of information from digital images. The information extracted from these images can then be used to perform a variety of functions. These functions can be performed simultaneously, thus eliminating the need to have multiple scanners that can only perform one function.
Figure 1: Machine visioning workflow
The 5 most common machine visioning functions
This function is most often used in one of 2 ways: Count the number of products, or the absence or presence of a product, component or feature.
Figure 2: a) Counting products b) Presence or Absence
This function can be used in a variety of different applications, here are some examples: Measuring the spark plug gap or the liquid level in a bottle.
Figure 3: a) Spark plug gap measurement b) Liquid level measurement
This function is most often used to decode information stored in a barcode, QR-code or OCR-code. The decoding function is usually used to track components as they moves through the production line.
Figure 4: a) Barcode b) QR-code
This is one of the most common techniques used to detect defects in a component. The components is compared to a defect free template image.
This is a very handy function and is often used in combination with one of the other functions discussed in this blog. This function can be used to locate a specific component in an assembly or it can be used to locate a specific reference point to position and orientate a coordinate system to take measurements.
Why make use of machine visioning
A machine visioning system saves money and increases profitability by:
If defects are not detected during the production phase of a product, it can cause costly recalls and damage the reputation of a company. Simple detection methods, such as verifying that the label on a bottle matches the content inside the bottle by comparing the label and barcode printed on the bottle, can dramatically reduce numbers of defective products. These types of comparison techniques are commonly used for medical products due to safety risks.
If defects are detected early in the production process, the defective component can be removed from the production line before it is implemented into a larger assembly. This avoids the possibility of scrapping expensive material and reworking/replacing parts.
Another way to increase the yield is to decreases downtime. A way to decrease downtime is to eliminate system jams by checking that a component is positioned and aligned correctly before it enters a machine. Misaligned components can be corrected or removed from the production line, thus eliminating miss feeds.
Track and trace part and products
Tracking work in process (WIP) is the ability to track and trace products during the manufacturing process. This helps the manufacturer to keep tabs on the components in a process, allowing parts to be made available just in time processing. This helps avoid component shortages reduces inventory requirements and shortens delivery time.
Comply with regulations
Complying with regulations is very important in the manufacturing industry. Insuring that a product is labelled correctly, the label is legible and contains all the necessary information to meet industry standards is very important. The ability of machine visioning to verify that all the information is correct and legible, makes this task easy to accomplish.
Key components of a vision system
Machine visioning systems come in many different configurations. Some systems can be very complex and others very simple, but the key components always remain the same. In this blog a brief overview of each component will be given. Time permitting a follow on blog will be written that will go into more details on each of the components.
It captures the image and presents it in the form of light to the camera sensor. When selecting a lens there are 4 important parameters to consider: Lens type, field of view, depth of focus and focal length.
Figure 5: C-mount camera lens
It converts the light captured by the lens into a digital image so that it can be processed. When selecting a camera there are 5 important parameters to consider for any given application: Camera type, resolution, framerate, interface and mono or colour.
Figure 6: Area scan camera
It is used to make the features of interest of a component stand out by maximising the contrast. The type and placement of the light is important to ensure features of interest are accentuated clearly.
Figure 7: a) Ring light b) Bar light, c) Back/front light
Software with various algorithms to extract information from images is used to perform image processing. The image processing component of machine visioning can be seen as the brains of the system. If the correct information cannot be extracted from the image it does not matter how well the other components integrate into the visioning system.
The basic steps of any image processing software are as follows: 1) Image Acquisition 2) Pre-processing 3) Analysis 4) Results
The typical result obtained from image processing software is a pass or fail, which can then be used to initiate an action using discreet I/O signals.
Other types of results can be obtained in the form of data, which can be used for tracking and verification purposes.
It is clear from the topics that have been discussed in this blog article that a machine visioning system can definitely help speed up a production process, but how effective that system will be is largely dependent on the component and software selection. To select appropriate hardware and software it is important that a client provide a detailed list of system requirements.