To understand the enormous progress that industrial image processing has made in recent years, it is helpful to take a brief look back. Today, as then, this development is closely linked to the development of PC technology. Initially, universal PCs were assembled into image processing systems by connecting individual discrete image processing components such as cameras, digital pre-processing of analog image data and special evaluation circuits. However, these systems had major performance limitations in terms of fast image acquisition and evaluation due to the slow parallel interfaces, and the distance between the test object and the camera was also limited.

Communication took place via traditional fieldbuses, which had limitations in terms of the available bandwidth. In addition, these systems were very much limited in their application to the fulfillment of their actual task; additional tasks such as comprehensive system monitoring were not possible. This was carried out by the classic evaluation computer via PC hardware monitoring, for example by monitoring the voltages on the motherboard or via HDD SMART. SMART monitoring is implemented in the hard disks themselves, monitors the status and reports the data and problems via the BIOS of the motherboard as soon as the function is activated. With the appropriate tools, the data can be permanently monitored so that impending failures can be detected in good time (at least some of them). This provides more security for the installation/data. In principle, an original industrial image processing system consisted of a camera head with a separate power supply, analog image transmission to the frame grabber, evaluation computer with connection to fieldbus/fieldbuses and other necessary system components.

Increased integration density

The development of increasingly powerful processors and robust, fast serial interfaces such as GigE Vision, USB 3.0 Vision and Camera Link, as well as the integration of these interfaces into the system components of the image processing systems, has significantly simplified the system structure. Further simplification was achieved with the option of also supplying power to remote system components via the Ethernet connection (PoE = Power over Ethernet). The power supply is achieved via the PSE (Power Sourcing Equipment), which supplies the necessary power to the consumers (PD = Powered Devices). The main advantage of PoE is that a power supply cable can be dispensed with, allowing Ethernet-connected devices to be installed in hard-to-reach places or in areas where many cables would interfere. The power supply to the device does not have to be supplied separately with a power cable and power supply unit or solved with a battery.

Instead, the device draws its power from the data network. In addition to the data signals, power must also be fed into the data line - usually at a central point in the network distributor. On the one hand, this can save installation costs, and on the other hand, the easy-to-implement use of a central uninterruptible power supply can increase the reliability of the connected devices. PoE is generally used by network devices that require little power.

A common system setup for image processing systems therefore consists of a powerful central computer, one or more digital camera heads with GigE Vision or USB 3.0 Vision interfaces and other components such as cables. The digital camera heads provide the data stream and can also be supplied with power via the fast interfaces. As fewer individual components are used here, both the system structure and the cabling are simplified. The cabling plays an important role here, as the range of the high-speed interfaces is quite long, which enables a much more flexible system setup with longer cables.

Eliminate weak points

However, any data transmission network is only as strong as its weakest link. Although the performance of a PC has improved significantly in recent years in terms of processor speed, memory access rates, bus speeds and architectures, the PC and the operating system must offer extreme performance under high load in an image processing system. This has led to the development of specialized image processing computers such as Efco's 'Eagle Eye' that are optimized for the task.

The basic system characteristics of the Eagle Eye systems are specially designed to meet the market requirements of industrial image processing. They focus in particular on the areas of factory automation, machine vision and transportation/logistics. For demanding image processing tasks, the Eagle-Eye systems have the latest high-performance Intel Skylake/Kabylake i3/i5/i7 CPUs and up to six Gigabit Ethernet interfaces, four of which are equipped with PoE-PSE function for supplying IP cameras. The two remaining Gigabit Ethernet interfaces can be used to integrate a real-time fieldbus such as Ethercat and also a classic TCP/IP network. A further four USB 3.0 interfaces are used to connect additional peripherals such as USB camera systems. Eight optically isolated inputs and eight optically isolated outputs are available for integration into the machine environment. The six classic serial interfaces, which include two RS485 interfaces, can be used to establish connections to older system components that are still in use.

System monitoring with AI

The Eagle Eyes series is designed for industrial image processing applications in terms of interfaces, motherboard, storage and power options.

© Efco Electronics

The use of artificial intelligence algorithms specially developed by Efco for evaluating operating parameters in the Eagle Eye Vision systems is a first for image processing systems. For this purpose, a separate ARM controller integrated into the device determines the system parameters and evaluates the measured values. If consistent trends in the change of measured values are detected, actions for predictive maintenance can be triggered or carried out as required. If the measured values reach a limit value defined and preset by the user, the ARM controller issues a warning to the application software or the operating software.

The resulting data can then be collected and statistically evaluated in a higher-level host computer or in the cloud. A small OLED display on the front of the device informs the user of the device status and displays the measured values or error messages. The user can then access the ARM controller and the display via an interface using the customer software.

Transmission via Sigfox

Another new feature is the ability to transmit status information via Sigfox using a special device option. Transmission is also possible if the computer system is no longer running or is switched off. Sigfox is a comprehensive wireless network that connects devices with low power requirements to the cloud.

The devices offer a wide range of interfaces, which can vary depending on requirements.

© Efco Electronics

The network is optimized for the transmission of short status information with low power requirements to the cloud. This means that battery-operated devices can also deliver data to the cloud over long periods of time. As the wireless maintenance uplink via Sigfox is available across the board, it also works in vehicles, for example.

Thanks to their features and wide range of functions, the Eagle Eye systems are suitable for many applications. For example, they can be used in factory automation to control and monitor robot arms or in industrial image processing for automatic optical inspection. Eagle Eye also performs these tasks in trains, construction machinery, police cars, cabs and buses, as well as in Google Maps' Street View vehicles. Unmanned vehicles, drones, ships and autonomous cars are also conceivable applications. These systems can also be used for surveillance tasks, such as facial recognition in intelligent buildings, public spaces and streets, as well as for intelligent lockers.

The Eagle-Eye series is divided into three versions - Entry-Level, Main-Stream and High-End - in order to always be able to offer a system with a suitable price for these diverse applications. These differ essentially in the performance of the CPU and the number of available interfaces. The entry-level systems have an Intel CPU from the Braswell class (Pentium, Celeron, Atom) as the computing unit. This results in a low power consumption of 6 W and very compact dimensions. Support for PoE as well as self-learning capability (LCM = learning content management) and a status warning display are standard in all versions. The mainstream systems are in the mid-range price segment and are characterized by the higher CPU performance of the Intel Sky Lake U processors i3/i5/i7 and a more extensive range of interfaces. This increases the power requirement to 15 W. The high-end systems are accordingly equipped with the high-performance CPUs of the Intel Skylake S series and a complete range of interfaces. If even higher performance or expansion is required, the systems of the Eagle Eye Pro series are available, which can be modularly tailored to the application with the appropriate plug-in cards.

Author: Helmut Artmeier is Managing Director of Efco Electronics.