While the rest of the world worries about credit crises stemming from sub-prime lending in the United States, prospects for the machine vision industry appear to be upbeat, if the mood of the European Machine Vision Association's (EMVA) business conference was anything to go by.
As ever the programme of the conference, held this year in Berlin, from 11 to 12 April, reflected its location to some degree. According to Gudrun Litzenberger, of the German Engineering Federation (VDMA), the VDMA's annual survey of the German machine vision industry has picked up discernable signs of change. Automation companies, such as Siemens, are opening up their own machine vision divisions and entering the market alongside the established players. And while inspection for quality control is still the main application of vision technology, the survey found strong growth in robot-vision applications, particularly in the large engineering companies.
About 29 per cent of the German machine vision industry's turnover came from automotive applications in 2007, but the industry is much less dependent on automotive than, say, the robotics industry, where about 50 per cent of turnover is tied up with car and vehicle manufacture.
She sees new customers and new markets developing for vision technology. Among them are pharmaceuticals and cosmetics; medical devices manufacturing; and the metal working industry. Outside of the manufacturing sector, there are growing opportunities in logistics and postal package sorting; in microscopy and life sciences; and traffic. This tour d'horizon of developing markets where the application of machine vision can spread out from its traditional industrial sectors gave her presentation an upbeat theme that countered her news that the VDMA's annual survey of the German industry recorded a revenue growth in 2007 of only three per cent – the lowest growth figure since 1995.
A constant theme of all machine vision events in recent years has been the opportunities for growth in the military and security applications of machine vision, and Gudrun Litzenberger listed this too as a developing area. However, it was not clear from the meeting to what extent there is either a technologically informed demand from the security sector or, indeed, a ready supply of appropriate technology from the machine vision industry.
Dr Hervé Borrion, from the Centre for Security and Crime Science at University College London, outlined for the meeting many of the CCTV surveillance initiatives that are being undertaken in the UK, but it appeared from his talk that much of this is proceeding without significant input from the traditional machine vision industry and that the quality of the images that are being produced remains very poor. Security applications have their own technical problems – uncontrolled ambient lighting is one issue, and the difficulties of image processing and automated facial recognition when the subjects are moving or facing in sub-optimal direction, are others.
In contrast to the technological under-development of security applications, the presentation from Dr Rob Giesen, of National Instruments, delivered a glimpse into a hugely sophisticated technological future. All machine vision companies are going to have to take account of changes in the design and manufacture of computer chips – both commodity and specialist processors. The development by Intel and AMD of multicore processors for commodity computers means that the emphasis has shifted from the 'number of instructions executed per second' to 'number of instructions per Watt' – power (and therefore heat dissipation) had become an important constraint. With multicore will come a shift toward parallel programming. But Dr Giesen expects that we will also see a growth in the use of commodity multicore chips in embedded applications, perhaps replacing some of the existing specialist processors.
He also suggested that graphics processing units (GPUs) will have an increasing role in vision applications. A GPU is a dedicated graphics rendering device for a personal computer, workstation, or game console. They are very efficient at manipulating and displaying computer graphics, and have a highly parallel structure. These properties will make them attractive for image processing purposes too, he suggested.
Although the EMVA conference was held in Berlin, one interesting aspect was the number of Italian delegates in the audience. According to Ignazio Piacentini, representing the Italian Machine Vision Group, the Italian machine vision industry ranked about fourth in Europe by turnover. However, statistics were difficult to gather, due to the fragmented nature of the industry where most of the companies had only a few employees. The Machine Vision Group counted 10 manufacturers, seven integrators and one university among its membership, but there were still significant players in the Italian industry, such as Tattile, who were not members. While the group has seven integrators, he reckoned that there were about 65 in Italy as a whole. In attempting the sum up the reasons why machine vision appears to be flourishing in Italy, he pointed to the fact that Italy excels at flexibly manufacturing and that flexible manufacturing needs robots and machine vision.
