Overcoming the Limitations of Vision Systems in Manufacturing

Advances in low-power CPUs such as the Intel Atom E3840 series have made it possible to integrate high resolution, fast frame rate and image processing functions into compact smart cameras, greatly improving application and cost of vision systems for manufacturing operations.



Suppliers in the world of manufacturing are continuously in demand of increased solution flexibility and productivity, reduced cost, and the ability to support adjusted production schedules – in essence, ever-advancing automated processes throughout the production chain. One area of automation that is helping suppliers meet their goals is automatic inspection by machine vision, which is leading the charge toward the “smart factory.” The conceptual smart factory is a place where errors resulting from human manual operations are all but eliminated, quality consistency is improved, productivity is increased, production costs are reduced, and customer satisfaction is improved.  With all of those boxes checked, operations in a smart factory clearly benefit from a distinct competitive advantage over the, well, less smart factory.

Time and money are always key factors defining that competitive advantage, and it is important for system implementers to recommend a machine vision system that effectively minimizes cost and time-to-implementation. Multiple solution options—with their own advantages and disadvantages—currently exist for machine vision applications. Embedded vision systems provide great computing performance, but with a larger footprint, more complex deployment, and higher price tag. Industrial smart cameras offer compact size and fanless operation, but require a lower power, lower performance ARM-based CPU with limited memory—and therefore limited imaging capabilities. However, the convergence of high performance and low power consumption on new processors has opened up the possibility for an industrial smart camera solution that offers the best features from both larger embedded systems and smaller conventional smart cameras, providing a new alternative for machine vision applications.

So, what are the essential requirements in the smart factory? And what types of vision systems could be selected to meet those requirements?

High efficiency and throughput are critical for the higher productivity most industrial manufacturers pursue. However, there is a cost. In terms of conventional machine vision systems, high resolution and high frame rate are hard to achieve at the same time.

Ruggedness and reliability are essential to the operating environments of industrial production, which are often challenging for automatic systems. For example, a food and beverage production facility is likely to present damp conditions with extreme temperatures, while machine tooling environments are often dusty with metal or other intrusive particulates present. If the vision system is to be installed adjacent to production equipment, a higher degree of imperviousness to such elements is needed.


Integration with third party equipment: A production line usually involves a series of operations from manufacturing, machining, pick and place, and inspection to packaging.  For instance, in computer numeric control (CNC) turning operations, a number of different machines are used with an external controller, such as conveyors or robotic arms to move components from machine to machine and align them under the guidance of industrial cameras before cutting operations commence. After turning, the objects are conveyed to the next operation stand for flaw inspection. Finally, approved products are sent to packaging and undergo barcode reading for shipping.  Integration of and communication among the different systems involved is a challenge for all smart factories (Figure 1).


Figure 1
A production line involves a series of operations that require communication between devices and systems in order to achieve a smooth, automated process.

Faster development of software solutions and related compatibility issues are critical factors, dictating success or failure of the implementation.  Shortening development time and reducing system development costs are distinct challenges.


Types of Vision Systems

To be a configurable vision system means to integrate with an industrial PC and different functional modules, such as motion controllers, frame grabbers, data acquisition modules, and serial communication cards, providing the best flexibility and choice. The configurable vision system could integrate with the latest server-grade CPU and high-bandwidth PCI Express gen 3 technology to deliver the highest computing power, compared with embedded vision systems and smart cameras. Such a system is generally used to connect to high-resolution industrial cameras, such as Coaxpress and CameraLink, to carry larger-sized images, or other systems that require high throughputs. However, the disadvantage of the configurable vision system is that it requires installation in a rackmount industrial chassis, which gives it a larger footprint.

Another type of vision system is the smart camera. Smart cameras are small, compact, all-in-one vision systems that incorporate lens, image sensors, system storage and processors into a single device, a combination of camera and computer. And the generally easy-to-use applications are included, which means the user may not need to have programming skills.Conventional smart cameras are often single-purposed and dedicated to simpler image tasking, such as gauging, counting, alignment, or barcode scanning.  Conventionally, smart cameras have made use of a low power ARM-based or single-core Intel Atom microprocessor with limited memory, in consideration of size and ruggedness. Due to their minimal expandability, realization of additional functions requires installation of more system units.


A third type of vision system, which can strike a balance between the capabilities and requirements of the smart camera and the configurable vision system, is the embedded vision system. Embedded vision systems are comprised of an industrial PC connected to high-resolution industrial cameras. Embedded vision systems typically feature a high-performance processor running a standard PC operating system with multiple vision channels supported to deliver a full set of image processing functions. See Table 1 for a comparison.


Table 1
Configurable vision systems, embedded vision systems, and smart cameras all offer benefits and pitfalls that will determine the best choice for a specific vision application.

Embedded vision systems are, however, often more costly and complicated to deploy. An increased footprint compared with a smart camera solution is also a disadvantage in often space-constrained production floors. The potential need for more cables and for fans also affects system reliability. In addition, embedded systems tend to be less rugged and are not ideal for harsh environments, such as production areas with humidity and dust.


Is it possible to find a solution that has it all?

In reality, vision applications are often a marriage of high resolution with lower frame rate, or lower resolution with higher frame rate. To achieve both, a more advanced CPU is needed with costs raised accordingly. Striking the delicate balance necessary among these factors and achieving optimal efficiency with reasonable cost structure is an important issue faced by system developers on an ongoing basis.

Intel’s introduction, in the fourth quarter of 2013, of the high compute performance, low power consumption Atom processor E3840 series has ushered in a new category of vision system that features small and compact all-in-one systems with full PC functionality, high resolution and high frame rate multitasking, flexible expandability, and easy deployment. The new generation x86 smart camera represents a combination of the advantages of the existing lower power, lower performance, smaller form factor ARM-based smart camera with the higher performance, larger footprint, costlier embedded vision system. The x86 smart camera offers a highly integrated, high performance, compact vision system based on Intel architecture—a new market trend.


Breaking the Boundaries of Smart camera and Embedded Vision Systems

The x86 smart camera defines a new category of vision system that singularly realizes high performance, maximum integration, easy deployment, space efficiency and minimal total cost of ownership, well beyond what conventional systems can achieve. New generation x86 smart cameras run on quad-core Intel Atom E3845 processors with serious improvements on CPU and GPU performance while conserving power expenditure.  The new processors provide the palm-sized x86 smart cameras with the combined advantages of both conventional smart cameras and embedded vision systems.

The FPGA co-processor and the GPU engine’s built-in CPU help to offload tasks from the main CPU, releasing resources for more advanced computing.  Multitasking is thus viable, allowing the x86 smart camera, though small in form factor, to simultaneously manage gauging, counting, alignment and 2D barcode reading operations.

We can see the improvement in a number of key factors. First, conventional smart cameras usually run on a single-core Atom processor or ARM-based processor with considerations for size, power and heat dissipation.  However, these conventional smart cameras have limited computing power and are often used only in simple image applications dealing with individual tasking of gauging, counting, alignment, or barcode scanning.

The new generation x86 smart camera equipped with the Intel Atom E3840 series has doubled performance over previous generation processors while retaining a power consumption under 10W, enabling multitasking in a fanless small system for the first time. Multitasking systems can reduce the number of installations, an economical advantage in terms of total cost of ownership (TCO).

Image sensors are the eyes of the vision system so larger sensors can acquire more image information and deliver higher image quality. In the past, with conventional smart cameras focused on simple imaging tasks, the size of image sensors was not an issue. However, with the implementation of high-end and high-speed applications, image sensor size becomes critical for image quality.


Higher performance enables the use of a global shutter. Rolling shutters and global shutters differ in the way their pixels collect light. Rolling shutters collect light in sequential rows, with each row starting and finishing collection slightly different from each other.  Global shuttering pixels start and end light collection during exactly the same period of time (Figure 2).


Figure 2
Global shuttering is the more efficient manner in which pixels collect light, but this method cannot generally be supported by ARM-based industrial smart cameras.

Conventional smart cameras, whose limited computing power is insufficient to process large amounts of image data, have tended to adopt rolling shutter function. Even so, the inability of rolling shutters to remove residual signals, such as blur/skew/wobble/partial exposure effects, when dealing with fast-moving objects has excluded conventional smart cameras from use in high-speed industrial applications. Currently, however, with the improved CPU efficiency of new generation Intel Atom processors, small form factor smart cameras are able to support global shutter deployment.

While image quality is critical for accurate automatic inspection and analysis, limits of optical conditions (light source or lens) frequently cause acquired images to exhibit inconsistent brightness, leading to misjudgment in analysis. If the vision system can automatically optimize acquired images before submission for analysis, accuracy of image analysis is significantly enhanced.

In conventional vision systems, captured image data is processed by the CPU. When processor resources are insufficient, the amount of image data able to be processed is reduced. Thus, conventional smart cameras must frequently compromise either image resolution or frame rate.

The use of an FPGA co-processor by new generation x86 smart cameras greatly improves image processing efficiency by offloading image matrix operations from the CPU to an FPGA (image pre-processing), freeing CPU resources to carry out more advanced algorithmic operations.  The FPGA co-processor can carry out image pre-processing tasks such as look up table (LUT), region of interest (ROI) and shading correction, with these smaller vision systems accordingly realizing faster and more complex applications.

The new generation Intel Atom E3840 processors adopt a GPU driver, which offloads media processing tasks from the CPU, tripling graphic processing performance over previous generation processors.  With built-in Intel HD Graphics 4000 technology, the GPU can process video encoding, compression and transmission across multiple channels simultaneously.  This performance improvement empowers small vision systems to record, store and analyze media data, resulting in a “smarter” factory.


The new generation also presents advantages in terms of display, instruction length and system storage. Conventional smart cameras transmit data only via an Ethernet cable connected to the control center. If the vision system can also connect with HMI or a screen at the production line via VGA or Ethernet port and display image data simultaneously, operators can view inspection results and find problems earlier. As image analysis applications are required to manage large amounts of data, most mainstream software tools in this segment utilize 64-bit instructions. System storage capacity can determine whether the vision system is able to run a full PC operating system and third party APIs in addition to the amount of image samples the system must store for matching and comparison.


Total Cost of Ownership

Total cost of ownership is not determined solely by the nominal price tag of the system, but rather a combination of factors, including space usage, peripheral support, system expandability and software development costs. The physical size of the vision system, including external cabling, should be considered as production space cost. External wiring and cabling, as well as extended peripherals such as PWM light source controllers, must also be taken into account.

The number of channels the vision system provides defines its expandability.  Conventional smart cameras, though cheaper in single unit price, present the need for more system units to accomplish necessary expansion, such that actual system costs are much higher.  New generation x86 smart camera systems provide multiple channels and a GigE port that supports an additional slave camera, obviating the requirement to install additional system units, reducing average channel expansion costs.

Another important factor is software development and versatility. As mentioned, a manufacturing facility comprises multiple operation stands, among which effective communication and integration determines actual factory efficiency. If existing software resources can easily migrate across systems, human resource and development costs in deployment are dramatically reduced.


The new x86 smart camera provides compatibility with GeniCam and GenTL standards for image acquisition, as well as an intermediate platform with the same API to communicate with 3rd party software (Figure 3).  New generation x86 smart camera systems can provide an I/O topology similar to those in embedded vision, including GigE, VGA, RS-232, USB ports and isolated digital I/O, maximizing communication and integration with other devices on the production line.


Figure 3
With versatile open architecture, programs written by developers can easily migrate to different devices, shortening software development time and reducing total cost of system expansion.


To develop a smart factory environment for modern mass production process, the implementation of automated inspection is crucial in guaranteeing manufacturing quality and productivity, a primary requirement in enhancement of corporate competiveness. Time and money are always key factors defining competiveness, and it is important for system implementers to choose a system that effectively minimizes cost and time-to-implementation. New generation x86 smart cameras define a new category of vision system that singularly realizes high-performance, maximum integration, easy deployment, space efficiency and minimal total cost of ownership, well beyond what conventional systems can achieve (Figure 4).


Figure 4
As one example of a new generation x86 smart camera, Adlink’s NEON-1040 features 4MP 60fps global shutter sensor and the Intel Atom quad-core 1.9 GHz processor, offering minimal footprint and rugged IP67-rated construction.


ADLINK Technology
San Jose, CA
(408) 360-0200