Smart Cameras in Life Science Applications
Smart cameras combine a camera with image processing and machine vision programs all in one package. They are commonly used in life science applications where there are space constraints or no room to mount a separate controller such as high end digital microscopes for off-line cell inspection, bar code reading for packaging and pharmacological products.
Vision sensors are considered to be lower-end smart cameras, but they have separate controllers. They can handle many of the same applications and perform similar simple tasks, and offer many advantages over a full-blown smart camera when the space exists for a separate controller.
Smart camera controllers are computers or microprocessors that process images in the camera itself instead of transferring them to a pc or separate processor. They can also be called intelligent cameras or embedded vision processors and are handy for customers who want the controller in one package. The smart camera industry has evolved over the last decade to have optics, lighting and a camera in the front end, and processor, communications, and digital I/O all in one compact unit.
They help minimize the complexity of the rest of the machine vision system, reducing the need for a separate pc or multiple illumination sources when multiple cameras are used. Using low cost smart cameras can also offer a price advantage in multiple camera systems effectively spreading the cost of the pc across multiple units.
Smart cameras are self-contained but are also used in multiple camera applications with separate controllers that allow them to perform separate tasks or work together. The controller acts as a single main brain controlling separate smart cameras. Vial inspections are one such application. One camera on top looks at the vial's cap and other cameras at different angles looks at the sides of the vial. These types of systems are look for presence/absence, or inspect the appearance to make sure there are no contaminants, determine color, intensity, perform width measurements or count objects or edges, or perform identifications reading 1D or 2D bar codes or using OCR tools.
In clinical or diagnostic instruments or in analytical chemistry applications where there is a PC dedicated to a user interface, smart cameras may not offer space advantages. But they offer the advantage of being able to offload some of the processing or validation functions off of the customer's software if reducing software complexity and separating functions is desired. Depending on what a user wants to do and the number of cameras needed, smart vs dumb camera and pc may be more or less advantageous.
Another reason life science customers use smart cameras is because they already have them on-hand. It may make sense to use smart cameras are already stocked for bar code reading and motion control, sometimes called imagers, on other applications.
Top Considerations for Life Science Users Considering Smart Cameras
The first consideration is whether the system is able to pick up the features and details it needs to. The camera should be able to perform the desired task. Some applications may involve reading color. Others may need to read codes or positions of labels or the depth of a position probe into devices.
Many times what can be done with a smart camera is well beyond what a particular application needs. Rather than buying smart cameras with features that are never used, inflating costs, it is important to consider what is actually needed, now and in the future.
There may be no reason to add color if the camera will be used to verify codes and look at position. Adding color typically doesn't add a lot of cost to the smart camera hardware, since price differences between monochrome and color cameras are not much different. But it may add a lot of unseen costs to the software or inspection tools, time for analyzing color, and unanticipated engineering time.
In any machine vision system a key consideration is why a smart camera would be used. Is it to reduce human error? Increase throughput? It is important to track and trace inspection data in the life science industry. Machine vision systems now have the ability to send data to an ftp server or output the imaging data to a pc. Tracking and tracing is done through software, but some smart cameras also have the ability to output this information.
The smart camera should also have the ability to output the desired data where it needs to go. Knowing the type of information that is to be output and type of connections needed helps determine whether simple smart cameras that perform simple tasks and have limited functions and limited I/O options, such as some sort of Ethernet, are suitable or if more complicated, powerful systems or separate controllers are needed.
Cameras Most Suitable for Life Science Applications
Resolution and package size the customer needs limits the type of smart camera that a life science customer can use. In many inspections, code reading, and 3D height mapping, looking at slopes or angles or thicknesses of items, such as cloth covered gel inspections on medical bandages, smaller cameras are better. Many products in life sciences are small, such as implantables, samples, slides, or tubes, and it is common to use high resolution smart cameras with a small footprint.
On the 3D side, used more in computer vision, there is more leeway in camera size because they are often not as restrictive. But the number of smart 3D cameras available to this industry is limited and few manufacturers supply them.
High resolution smart cameras aren't always needed. Color versions use a lot of processing power to handle the color information and generate a lot of heat. Smart cameras are available in a wide range of resolutions so they are able to resolve or pick up the feature customers are looking for.
Color or Monochrome?
Monochrome smart cameras are typically used whenever accurate measurements of products are needed. They offer higher resolution images since the pixels don't have filters on them to extract color information, and they offer slightly faster imaging processing speeds. They are usually slightly less expensive than color cameras, but not by much.
Color smart cameras can extract features that aren't possible to extract with monochrome cameras, and are relatively quick. They are used in ID reading and inspection applications in life science industries. In many cases, color smart cameras with monochrome tools offer the benefits of being able to read contrast information without having to use true color. This type of setup offers flexibility for the future to customers In life science industries who need to 'replace exact' after an inspection system has been developed, approved, and frozen, as opposed to other industries that allow compatible parts to be swapped.
Smart camera manufacturers control the life cycle of a particular smart camera based on the availability of its specialty sensors. If they rely on commercial items, sensor types may change faster than the machine is allowed to change.
Digital or Analog?
Most life science applications use digital smart cameras. Analog output is primarily used in applications that require continuous feedback, like tracking the edge of something or feedback for position to adjust the speed of a system, but there are not many analog needs in vision systems in life science industries.
Analog smart cameras are more often used in the measurement world for monitoring and when there is a human interfacing with a machine. But digital smart cameras are taking over inspections and monitoring processes with their better speed and resolution options. The cost differential is also reducing, so digital will soon be the preferred choice for monitoring applications as well.
CMOS or CCD - Which Sensor Type?
In the past, CMOS technology was inferior and cost less, and was used often in lower-end smart cameras. But the technology has evolved to the point that performance with CMOS sensors has become equivalent to CCD sensors at higher speed. Many new smart cameras use CMOS today. As long as the camera offers the resolution and speed the customer desires, the sensor type doesn't generally matter.
In the installed smart camera base there is a good combination of both. As more chip suppliers abandon CCD creation, almost every option in the smart camera market will be controlled by CMOS. Prices have come down due to their proliferate use in smart phones and other industries, and CMOS chip production is high, so CMOS sensors are very competitive compared to CCDs which are still used in only some applications.
Smart Camera Costs
There is a wide range in prices for smart cameras. Simple vision sensors cost around $1000-$2000 for basic systems. There are cameras and vision sensors in the grey zone around $2000-$3000 range with flexibility to solve several applications, but most smart cameras start around $4000 and go up from there.
Lighting, lenses, and resolution dictate the final price of a smart camera system. A standard 640x480 camera and basic light setup can cost around $5000. High resolution systems around 5 to 21 megapixels can cost $20,000-$30,000 depending on the lighting and lenses.
Pricing for smart cameras, bar code readers, and laser-based measurement devices in the life sciences industry is competitive. But, the automotive industry enjoys even more competitive pricing because this technology is more heavily used on production lines and manufacturers and vendors can offer deeper discounts. Usage in the life sciences market is more image-based, and smart cameras don't only read codes but they are able to provide multiple types of information about products. So fewer smart cameras are required, meaning prices haven't yet been impacted due to volume.
Main Challenges
The main goal of any vision system is to light the target correctly and obtain the needed contrast. This can be a challenge if there is little color difference, or when imaging curved or shiny parts that are difficult to light, which is often the case in the life sciences industry making lighting the number one obstacle for vendors who are asked to solve problems with smart cameras. Vendors offer lighting systems to be used with smart cameras that light the target from different angles and create shadows to highlight anything with a different height, like a scratch.
Another challenge in the life science industry for smart cameras is keeping personal data secure. A very small object with a bar code now can contain a lot of personal information that may be protected depending on various health care or privacy laws. How the personal data associated with a measurement is stored and transmitted, and whether there are data lags is becoming an important challenge. Smart cameras are used to take an image and prove that something is present or missing. The amount of data in a bar code for may become questionable for users who submit blood samples, and how to handle the timeline that the data is saved before it is stored is a growing concern in the customers' eyes. They may not want the data stored in different places and only want it stored on their drive.
Fastest Growing Life Science Application
Bar code reading continues to be the fastest growing application for smart cameras and vision sensors in life science, whether on the industrial or vision side of the market. There are many applications that straddle the line between what's required in a vision system. Systems dedicated to only bar code reading or only imaging are no longer needed.
Vision sensors now come with tools that are looked at as smart camera tools. They help vision sensor systems and inspection tools find, rotate, and read 1- and 2-dimensional bar codes on difficult surfaces, read very small codes, or many codes at the same time. They validate position, read text off of labels, fill levels on vials, placement on stopper caps and look at an entire tray of blood samples reading 9 codes at once to provide the code readout and potentially the position of the code so the machine knows if any are missing. This area shows a lot of opportunity for both smart cameras and vision sensors. Often, vendor are able to tailor the toolset options on a vision sensor system to be able to provide 90% of the customer's requirements that is easier to use and a lower cost option than a full smart camera.