|
|
|
Machine Vision Meets Digital Age Digital camera could return machine vision to a more prominent place in the market; automated inspection also has a shot In the 1980s, machine vision, "emulating human vision, had a lot of sex appeal," says Patrick Costa, president/CEO and chairman of Robotic Vision Systems, Inc. (RVSI), the market leader in semiconductor lead inspection. Machine vision was at the forefront of a whole robotics thrust as companies raced to invest in the high-profile endeavor. By 1986-87, however, this infatuation had vanished as investors who underestimated the capabilities and limitations turned their attentions to other ventures. Consequently, many companies were shuttered or were folded into each other. For instance, RVSI would acquire two other pioneering machine vision firms, Automatix and Itran, which had previously merged together. Now it appears the digital camera could return machine vision to a more prominent place in the market. Automated inspection, another area of machine vision, stands poised to enter the digital camera age as well. Richard Amtower, president/CEO of CR Technology, the Laguna Niguel, Calif.-based provider of advanced optical and X-ray inspection systems for semiconductor and printed circuit board (PCB) manufacturers, sees the digital camera making a strong impact. He notes that CCD cameras, the mainstay of machine vision, were cheap, reliable and did the job. At first, the digital camera could not compare. The megapixel cameras necessary for machine vision were not around several years ago. And when megapixel cameras entered the market, prices ran from $20,000-$40,000 and upwards. Based on the industry's mantra of 'better smaller, faster, cheaper,' the issue in using digital cameras was never about whether they add a lot of value, but what is the cost. "If you are looking at really accurate image dissection or analysis, having it digitized in a camera rather than a frame-grabber means lower noise, which relates directly to sensitivity, which in our case relates directly to accuracy," says Mr. Amtower. Most conventional machine vision systems deal in 8 bits or 256 gray levels. According to Mr. Amtower, digital cameras seem to be able to deliver better information in terms of gray level and with greater sensitivity and repeatability, i.e., more accurately. The advantages of digital cameras do not end at the resolution of the image, according to Mr. Costa. "You can add intelligence to it, changing the image and modifying it in real time, enhance contrast, color, all that stuff you traditionally did in the darkroom. You can download that and play with it endlessly." The issue in bringing machine vision into the digital camera age was largely of cost. Software intelligence is always going to be a significant portion of a machine vision system's cost. Mr. Costa's analogy is the relationship between the eye and the brain which team up to give us sight. The eye is the camera, but the brain interprets that image. "It was always true that the value added was our software -- half of the cost is software and rising. The brain has always been the dominant issue of machine vision," says Mr. Costa. Consequently, in order to reach an aggressive price point, given that the software expense was increasing, was to find an eye that was accurate and inexpensive. Initially, digital cameras did not meet this test. "That's sort of swung around recently in the right direction. It is moving where the cost is now coming into a reasonable range, under $10,000. Certainly, we see it moving in the cost/performance curve, not quite Moore's Law, but significantly," says Mr. Amtower. What is enticing about the digital camera is that its suppliers are targeting the consumer market while looking to sell in the millions, meaning costs will come down dramatically. "Now when you have millions of consumers that might buy digital cameras, the cost and performance of these sensors improve dramatically," says Mr. Costa, who adds "we, in machine vision, get to ride that curve down." "We think the price performance trade-off is favorable enough now, so that we are releasing them in products early next year," says Mr. Amtower. The combination of a dramatic cost and performance improvement in the eye combine with further advances to the brain, the machine vision software, prompting some to believe machine vision may deliver on the promises of the 1980s. "It allows you to get to new segments," says Mr. Costa. Machine vision is a $2 billion-plus market today, but Mr. Costa believes, "We are just scratching the surface for machine vision. I think these are just a few percentage points of the market of the future." In developing further applications for digital camera-enabled machine vision, RVSI has a joint venture with Polaroid. Mr. Costa says product announcements are forthcoming this year. In terms of branching out, RVSI believes bar code and machine vision technologies will converge. They will be promoting the 2-D bar code developed by RVSI's CiMatrix division, which recently was consolidated with Acuity, RVSI's 2-D machine vision company. RVSI Acuity CiMatrix will be headquartered in Canton, Mass., and will be led by Peter Cole, president of the former CiMatrix division. Mr. Amtower believes the digital camera will also shake up the status quo in the machine vision world. "The fact that the cameras are coming down in price and can do more, I think there is going to be more access to this technology and I think that means the high-end companies are going to lose their advantage. I think the more widely available a technology is, the more the customers are benefiting, because more suppliers can be involved in supplying this," says Mr. Amtower. |