Automation is becoming a software industry. As such we are seeing the beginning of a proliferation of software for automation components. Vision is just a subset of the general automation industry. To date, no vision software supplier is making money by only selling software. In fact, this is true for automation in general - no automation company is making money with software only products. Many companies are banking on the fact that this will change. The successful machine vision suppliers all focus on market niches, where the products they sell address specific applications. The most successful machine vision companies have aligned themselves with the semi conductor industry.

The vision market is still small. To achieve growth we are seeing an increasing number of acquisitions, e.g. RVSI acquiring Acuity, Cognex acquiring Isys, Komatsu etc…, ESI acquiring AISI, Orbotech acquiring Schuh are just a few. Companies are continually entering and leaving the market. Intel and Microsoft are developing vision software which is available free of charge.

The machine vision suppliers fall into three categories, hardware and software component manufacturers, system integrators and finally in-house developers. Many large machine vision users have developed their own hardware and software for their specific applications.

Trends
Both Microsoft and Intel are targeting the home consumer market. The implications for image processing has been the evolution of MMX, digital cameras, Firewire i.e. fast digital networks for audio and video. These companies will, and are, introducing standards for image capture, storage and processing. By the nature of the size of the home consumer market this drives some of the machine vision market.

High Definition TV will see the introduction of higher resolution cameras at a lower cost. Intel has announced PC camera designs and interfaces. Digital cameras are already storing images and doing some of the preprocessing.

Processing power doubles every 18 months, retailing for the same price and this is likely to continue for approximately 10 years.

Color standards are evolving with calibration templates to ensure the color matching for the printing process.

With the emergence of BGA and bump wafers 3D image capture and x-ray image capture is gaining momentum.

To date, the vision suppliers have been concentrating on tools for the OEM market. There is a growing trend to build reconfigurable vision systems for the end user market. Examples include Wit from Coreco, LabView from National Instruments and Checkpoint from Cognex.

Patents are taking a more dominant role in vision technologies. Nearly all major suppliers are pursuing patents. Having said this, I would also say that the basic image processing tools have not fundamentally changed. The list includes:

Image filtering
Image/Camera Calibration
Image Histogram Analysis
Blob/Connectivity
Edge/Caliper Tools and their Geometric Derivative
Contour Finding
Hough Vectorization
Correlation/Search/Pattern Finder
Auto Focus
Template Matching
Texture Analysis
Optical Character Verification / Optical Character Reading
Illumination Correction
Color Measurement
Image Mapping functions
Segmentation
3D Image Capture

Conclusions
There is no one vision supplier who has a monopoly on the best vision tool kit. The focus of different suppliers is being dictated by their revenue stream. To a large extent machine vision is a commodity item.

At this point machine vision cannot match what the human eye can do, so in this light we are still in the infancy of its potential.

Experience tells me the ingredients of a successful company are Imagination, Inspiration and Dedication, guided with Direction.

By:

Frank Meyer

Frank Meyer director of Soft Automation. Previously he co founded Integral Vision and did machine vision research at Cranfield Institute of Technology UK