We started our discussion by exploring the components of an industrial vision system. “A machine vision system has five key components that can be configured either as separate components, or integrated to a single smart camera,” explained Callejon. “The correct configuration is going to depend on the application the complexity of that application, but of the five key components you're normally going to find, the first one is going to be lighting. This is a critical aspect of machine vision to illuminate the part to be inspected, allowing its features to stand out. The second component will be lenses, where the image is captured and presented to the sensor. The third component will be the image sensor that converts the digital image for the processor to analyze. The next component is vision processing, which consists of algorithms that review the image and extract the required information for an inspection. And the last component—which is optional—is where the resulting data will be communicated to outside of the camera.”

Deriving from these components are four categories of industrial vision technology, Callejon said:

• Smart cameras—a combination of all the aforementioned vision system components into a single device;
• Vision systems—where image acquisition is separate from the vision processing controller;
• PC-based cameras—where the vision application software is installed on the user's PC or industrial PC connected to a camera via Ethernet; and
• Industrial cameras—which allow the user to visualize images with third-party software or with their own software application to inspect that image on a PC.

Explaining how to decide among the different vision technologies available for inspection applications, Callejon said to first consider the needs of the inspection application. “You need to clearly define what is an acceptable product and what isn't. So set clear judgment criteria, because you're going to use that when you're programming your devices.

Another important aspect to consider is: Will the inspected object be moving or stationary? Knowing this will help determine the type of image sensors you want to use. Also think about the space available to install the camera on the machine. Finally, consider the available budget, both in terms of hardware and service. This matters because, if you design the vision system entirely from components, you could save some of the integration and service costs. But you need to check if your team has the skills to work effectively with that kind of machine vision application.”

When it comes to inspecting either solids or liquids, Callejon said there is little difference from the hardware and software points of view. “The same vision system tools will be used, they’re just used in different ways. This is most evident with lighting. For example, a transparent liquid level inspection would normally use a backlight. While solid, opaque label inspections would use a more direct lighting approach.”

Motion also affects vision system choices in liquids and solids inspections. “For liquids, a smooth acceleration and constant speed or velocity will help to maintain level and consistent product separation,” Callejon said. “For solids, the maintenance of speeds is not as critical as it does not tend to change the characteristics of the item to be inspected.”

As with any automation technology, justifying the investment is always a critical step in the process. Callejon says the process for determining ROI (return on investment) from a vision system will likely differ depending on the industry. However, there are five common aspects to consider:

• Repeatability of inspection and consistent judgments;
• Speed of inspection to avoid bottlenecks on the manufacturing line;
• Net content control and waste management—this is key to maintaining labeled amounts to avoid penalties and waste;
• Labor costs—think about how you could repurpose inspection operators into higher-level, value-added activities once you have a vision system for inspection. This is particularly important today in light of COVID-19, as the use of vision systems can help reduce human contact; and
• Initial capital equipment costs and engineering costs.

Klebe noted that users should also consider other vision system benefits, such as the speed of identifying the root cause of an issue. “The speed of vision system inspection results and the ability to aggregate the data or use visualization software allows the manufacturer to much more quickly pinpoint the cause of a manufacturing problem and get the system improved or tuned so that they're producing quality product again more quickly,” he said. “These kinds of savings aren't often considered on the front end of a project. But if you look at the cost of downtime and the expense of poor quality in terms of scrap and rework, for example, all those factors should all be considered as benefits when you're doing your return on investment calculation.”