Industrial image processing has long been established as the standard tool for quality assurance. Two trends are now driving its development ahead: the accelerating pace of automated production, and the increase in quality requirements. The critical factor in meeting these challenges is software that can keep up with developments in optical sensor technology, cameras, interfaces and hardware.
To meet these challenges, The Kistler Group offers its KiVision: innovative image processing software that features user-friendly tools and optimized algorithms. As an added benefit, KiVision can also integrate customers’ individual new requirements as well as highly complex application-specific routines.
The phenomenon we are now seeing is certainly not a new one: any provider who offers a solution that simplifies difficult tasks or reduces the time required for them will inevitably trigger new requirements and expectations. The triumphant progress of email, for instance, has generated sky-high expectations for response times from recipients, with an enormous increase in the volume of messages sent. These developments are prompting email users to demand efficient filters and other helpful functionalities.
Industrial image processing has experienced similar developments in recent decades: there has been significant progress in measuring and testing series parts including turned, punched, pressed and plastic components. Automatic image processing systems are now in widespread use for these applications, and they represent a major success factor for production automation and quality assurance in particular. Thanks to their speed and cost-efficiency, for example, they allow 100% inspection which – unlike statistical monitoring – can also detect sporadic faults. At the same time, these systems are vastly superior to costly manual testing methods that are vulnerable to errors. These are critical advantages in view of the growth in quantities manufactured by industrial producers and the sharp increase in quality requirements. Such trends mean that manufacturers need high-performance, high-precision automatic image processing systems if they are to keep up with developments in automated production and protect themselves against claims based on complaints.
Users are becoming largely autonomous
The constant increase in requirements for industrial image processing presents a huge challenge for the software. For instance, high-performance automated test systems from Kistler deploy as many as eight cameras that generate ever greater quantities of data. “As a key component of industrial image processing, the software has to play its part in overcoming these obstacles – and it has the ability to do just that,” according to Ferenc Toth, Head of Kistler Vision Systems in the Kistler Group. Thanks to optimized algorithms, high-performance image processing software such as KiVision can cope with the breathtaking increase in series production throughput rates. Today’s automatic punching presses, for example, quite often operate at frequencies of 1 800 strokes per minute – which means that only about 30 milliseconds are left to capture and evaluate the images. By way of comparison: only a few years ago, press frequencies of 1 000 strokes per minute were greeted with amazement. Advances have also been achieved in terms of accuracy. Precise measurement of series parts with transmitted light inspection has now become standard, and accuracy requirements in the micrometer range can be implemented here with no problems.
“At the end of the day, this means that customers who have the appropriate systems and software are now independently able to perform many of the tasks that were still reserved for specialized industrial image processing providers just a few years ago,” Toth explains. “In spite of – or actually, because of – these achievements, expectations of the performance and operability of new software are very high – and the focus is now shifting onto more complex tasks. As we continue to develop our software, our claim and our objective will be to simplify the difficult problems faced by users to such a degree that they can eventually accomplish the tasks themselves, just as if they were standard procedures,” Toth adds. For surface inspection with reflected light, achieving this goal involves a particular challenge: complex algorithms must be translated into software that is easy to interpret and simple to operate. As an example: it might be necessary to identify extremely small scratches on sheet metal parts but, on the other hand, to disregard flaws such as blemishes on the fittings if they are not relevant for the quality of the manufactured parts. “In situations such as these, the focus is not only on the software as the control center but also on optimal interaction between illumination, optics, camera, interfaces and the computer,” Toth notes.
Integrated, straightforward and extendable
KiVision image processing software is integrated into all Kistler testing stations, but it is also used outside of test automation and can be combined with other measuring systems from Kistler. KiVision is Windows-based; just a few hours of training will enable operators on the factory floor to use the software independently and carry out their first measurement tasks. Aided by a comfortable, intuitive user interface, a wide choice of well-structured test commands and visualization of parameters, new users can even program testing tasks for entirely new parts with no outside support. Complex testing sequences (which mainly occur in connection with surface inspection) can be made much simpler thanks to reusable subprograms and function blocks. The software includes an innovative, integrated front-end designer that allows all users to assemble their own test displays as appropriate to their specific applications, with the help of the Drag & Drop function. Users themselves decide which camera images, measurement values or counters they want to see on the display – with no need to pay for costly special developments.
KiVision also supports all major camera interfaces and fieldbuses, and can be extended without problems. For example: customers can use the KiVision Software Development Kit to integrate special sensors or their own algorithms, with no need to disclose proprietary algorithms – a critical advantage, especially if patent applications are planned. Data transmission to external CAQ systems and integration into the customer’s own corporate network are also possible. These features allow external test programming at a PC workstation as well as central management of program data and measured values. In response to customers’ requests, additional software modules can be integrated to enable remote diagnosis and remote maintenance of automatic testing and sorting systems.
Fast and reliable in every situation
Modern high-performance processors in PCs contain ever-increasing numbers of CPU cores for ultra-fast processing of large data volumes. KiVision’s basic architecture is designed for parallel access to the power of the cores so as to utilize increased PC performance for faster image processing.
One essential factor in quality testing of safety-critical parts is reliable separation of bad parts – and this must also be ensured in exceptional situations. For this reason, Kistler accords the highest priority to guaranteeing that its automated test systems will prevent faulty parts from ending up in the OK parts container under any circumstances – during normal operation and also in numerous abnormal situations – and at the same time, there must be no loss of speed. To give customers the certainty they need, Kistler ensures smooth interaction with a well-programmed PLC, asynchronous handling and KiVision, which supports asynchronous handling; for example, virtual part numbers help to assign related images to the correct test specimen even if they are captured at different times. For this purpose, the PLC (as the primary instance) collects the results from the individual tests, and it can make a reliable good/bad decision or even stop the machine if need be.
Eager for more progress
Despite all that has been achieved in industrial image processing, the pace of innovation continues unabated. Kistler aims to offer solutions for the constant flow of new and more extensive requirements from customers, so the company is continuing to develop its KiVision software at the Vision Competence Center in Karlsruhe, and to offer comprehensive training. The image processing specialists at Karlsruhe collaborate closely with Kistler’s teams in Straubenhardt: their vast stock of industry expertise helps to expand KiVision by adding routines for specific customers and applications. The software is part of a coordinated complete package supplied from one single source: expertise in mechanical engineering, image processing and software as well as individual feasibility analyses and intensive dialog with customers – a combination that generates impetus to develop new solutions. Toth concludes with a summary: “We are making use of our concentrated expertise in optical sensor technology and image processing to offer innovative, user-friendly software that will give customers even more autonomy and open up yet more possibilities for them. Although so much has already been achieved, the vast potential of industrial image processing has not yet been exhausted, especially with Industry 4.0 in view.”
Tel: +41 52 224 11 11