Digital vision is a relatively new technology that has advanced dramatically in the last 10 years. The latest generation of vision systems has been dominated by the rapid evolution of smart cameras, which, in many cases, combine the intelligence of PC with the image quality of a high-end industrial camera. This article will explore current and future trends in smart camera technology. It also will discuss smart camera benefits and limitations compared to PC-based vision systems.
Definition of "Smart Cameras"
A smart camera, or “intelligent camera,” is a self-contained, standalone vision system with a built-in image sensor along with the processor and, in many cases, an FPGA required for image processing and communication. It contains all necessary communication interfaces (e.g. Ethernet), as well as 24V I/O lines for connection to a PLC, actuators, relays or pneumatic valves.
In general, smart cameras use a defined tool set to tackle various tasks that include, but are not limited to, presence detection, measurement, alignment, pattern recognition, ID reading (1-D and 2-D bar code reading), surface inspection, OCR/OCV (optical character recognition and optical character verification) and robotic guidance. These types of applications are found everywhere in the modern manufacturing, such as verifying that all the items are in a box and reading a 2-D data matrix code and a printed date code. Many other applications center on the quality of parts in production and the measurement and verification of components so they meet specifications. These range from the selection of the correct tie rod in the automotive industry to the detection of edge defects and stress marks during the fabrication of solar wafers.
More Widely Used
Smart cameras are being used by a wider group of people. The camera’s simple application development tools enable it to be used in more general machine vision applications. They are relatively easy to deploy because their size continues to get smaller, allowing them to be used nearly anywhere in the factory. The relative ease, with which these types of solutions can be developed, combined with the low price point, makes smart cameras the perfect fit for end-user and one-off applications. When compared to other types of vision systems or imaging devices such as frame grabbers and PC-based vision systems, some users see limitations in processing power and tool design. Other expert users still prefer the flexibility that the PC-based system can offer. Users can add more processing power relatively easily. They can also use standard software development languages like C/C++ and Visual Basic to create specific tools to solve applications. These specific tools often provide excellent and elegant solutions, but the development of these solutions can take many hours of programming and testing often making the cost of these solutions much higher especially for single installations.
Smart Camera Trends-New Applications and Technology
There are growing trends in smart camera technology, including advancements in the CPUs used in cameras, the specific processing devices used to improve image processing, operating systems and interfaces that make cameras more capable than ever before. Another development in the advancement of smart cameras is the addition of 3-D scanning technologies into this product range. Advancements in computing power along with the use of specific processing devices like FPGAs have made the intense processing needed to handle the immense amount of data produced by a 3-D camera viable within a smart camera package. The FPGA helps reduce the data quickly to a useful and unique 3-D shape and then the processor does the rest of the work of the application.
The increased power and flexibility of smart cameras have opened up several new application areas and even new markets. Applications that would have required the additional processing power of a PC- based solution, like high-speed OCR and even 3-D scanning are now handled very capably by smart cameras. As far as industries that previously were only served by PC-based systems, like biomedical imaging and even surveillance, are now being evaluated using smart cameras.
In the biomedical industry, applications such as indentifying interesting candidates for future investigation of cell clusters that show the likelihood of cancer or other potential disease allows medical technicians to evaluate greater volumes of material to validate test results. Also, these cameras can help reduce the time it takes patients to get their results. In most cases, the camera itself is not used to make a diagnosis but to increase the repeatability of these tests throughout the industry.
Similar to the biomedical industry, surveillance applications can require sifting through mountains of data to track and highlight interesting suspects for future and continuous investigation. Cameras can be programmed to track and monitor individuals moving much faster or slower than the rest of the monitored population, for instance. The cameras also can be used to document or catalog people to see if the same person is returning to the area multiple times, possibly casing the area for future actions. Smart cameras can be used in conjunction with one another to track suspects once they are identified as they move from monitored area to area.
These applications are becoming more and more viable using smart cameras by using different and slightly more open architectures and allowing users to port applications to new cameras when major advances are made in camera resolution, or processing power becomes available. In traditional smart cameras, the units run specific operating systems to allow access to the proprietary set of tools and communications functions. The newest trend in the smart camera is the PC within the camera running a standard computer operating system, like Windows or Linux. By running these operating systems, users are able to create solutions using either higher functioning vision libraries (several available from multiple vendors and some open source options). There are even versions that allow the user to deploy solutions created in standard programming languages like C++. These new versions of smart cameras, or “super-genius” cameras, still fit the definition of a “smart camera” because they are still self-contained devices containing imager and processor that can be deployed for a specific application. Manufacturers have started to blur the lines between the PC vision-based solutions and the traditional smart cameras.
Super-Genius Smart Cameras
In the past, it was easy to categorize machine vision systems: there were PC-based systems and there were smart cameras. Today, smart cameras have evolved a sub-genus, the vision sensor. And at the top end of the smart camera spectrum is a “super-genius” group that has been slowly emerging; the PC-based smart camera.
The vision sensor end of the spectrum has also gone through a massive growth spurt. There have been a huge number of new entrants into the market with vision based solutions. These products are similar to smart cameras since they have an imager, optics and processing in a self-contained housing. The tool set that these devices possess is typically a subset of the tools from a smart camera.
Vision sensors are even easier to use and deploy. The software is designed to be very intuitive for even non-vision users. The vision sensors category is typically lower cost and more specialized in their application solutions. To keep the cost down, vision sensors are often specialized for applications like inspection, positioning or code reading. The one thing that does not seem to be reduced in this category is the communication method. They may not be fully customizable like they are in the PC and smart camera versions, but they offer a lot of flexibility in the low cost category.
The overall trends in the smart camera market include not only more processing power and more flexibility in tool set and tool creation, but also specific application solutions at a lower cost called vision sensors. This greatly expands the market for smart cameras as they become suitable for more and more applications. They have more functionality on the top end with the “super genius” cameras and more accessibility on the low end of the market, reaching applications that may have been priced out in the past. The only thing that seems constant in the smart camera market is change. Look for more power and flexibility on the top end and more easy-to-use functions throughout the spectrum, making vision solutions truly smart.
Users can add more processing power relatively easily.
They can also use standard software development languages like C/C++ and Visual Basic to create specific tools to solve applications.
Smart cameras using different and slightly more open architectures allow users to port applications to new cameras when major advances are made in camera resolution, or processing power becomes available.