Market growth demands customer focus
Andy Wilson Editor at Large
According to a recent report by the Automated Imaging Association (AIA), the North American machine-vision market exceeded $1.3 billion in 1996. Nello Zeuch of Vision Systems International (Yardley, PA), the report author, claims that approximately $800 million is from sales of application-specific machine-vision systems, about $200 million from general-purpose machine-vision products, and some $300 million from miscellaneous image-processing boards and services.
The market study also predicts that by the year 2000 the machine-vision market will climb past $2 billion. With such large numbers, one would expect that sales of OEM products, such as cameras, boards, and attached processors, would also be prominent. Even if this is correct, OEM suppliers are not likely to double their business in the next four years. Indeed, companies in the add-in board market seem to have a habit of stalling after reaching $20 million per year in sales. Although the reasons may be business-related, a more reasonable explanation lies within the technology itself.
Although several OEM manufacturers, such as Imagenation (Beaverton, OR) and Data Translation (Marlboro, MA), sell many frame-grabber boards, their profit margin on such low-cost products is small. Supported by driver and third-party software products, frame grabbers rely on the host central processing unit to perform image processing. Though extremely useful to developers of microscopes and other nonreal-time image-processing systems, dumb frame-grabber boards are often unsuitable for high-speed machine-vision systems.
To reach a larger share of the machine-vision market, companies such as Datacube (Danvers, MA) and Imaging Technology (Bedford, MA) offer sophisticated pipelined or digital-signal-processing products that can accelerate image-processing operations. However, although the profit margin made on such products is greater, the market is smaller than less-complicated products and requires more software support. Worse still (for the OEM), customers for high-end products tend to be more knowledgeable, thereby demanding a higher level of engineering support.
Most successful companies
Interestingly, the most successful companies in the machine-vision market are those that build application-specific products, as the research study numbers from the AIA suggest. Companies such as Cognex (Natick, MA) do not simply develop add-in processors and software tool kits. Rather, they have focused on a specific part of the machine-vision market and have chosen to tackle a larger part of the end-user`s problem. Accordingly, such companies can demand higher prices for their hardware and software. And once tackled, such solutions can be applied to other customers with similar problems.
One of the main reasons for the success of machine-vision-related companies is what appears to be a shortage of systems integrators who understand enough about specific areas of machine vision. Unfortunately, as anyone who has been involved with image processing knows, there is no easy way to make money in machine vision. Every application demands a specific set of engineering requirements. In many applications, off-the-shelf image processors cannot be used, and solutions must be tailored for each individual system. Tackling such problems demands a close relationship between OEM supplier and customer. Hopefully, the growing machine-vision market will foster more research and development in image processing so that application-specific machine-vision systems can be built more easily.