Contrast and
compare
Common application
programming interfaces
will speed time to market.
For the most part, the machine-vision industry has com- years on finalizing the GigE
prised a relatively small number of companies with turn- standard for cameras.
overs in the $20 million/year range. Consisting of sensor, The GigE standard would
camera, lighting, frame grabber, and software vendors, these result in cameras that are
companies have been successful because they offer very Internet Protocol devices,
specialized products and the technical expertise to support make good use of existing
them. In many cases, end-users in industries such as phar- IP standards, and serve
maceuticals, electronics, or lumber have been faced with the machine-vision market
building a high-speed product inspection system but have well. Many machine-vision
no expertise in machine vision and so must contact these companies have announced
vendors or their system integrators to retrofit vision into an or plan to announce GigE-
automated assembly line, an often challenging task. based products. Pleora Technologies (Ottawa, Canada), a
For machine vision to become more widely accepted in pioneer in developing GigE, has launched an OEM board for
the industrial-automation world, this scenario will have to in-camera GigE connectivity, and has agreements with cam-
change. With the time to market of today’s systems being a era makers such as Atmel (St. Egreve, France), Basler Vision
fraction of what it was 10 year’s ago, system integrators need Components (Ahrensburg, Germany), Dalsa (Waterloo,
to rapidly evaluate and compare cameras, frame grabbers, Canada), and JAI Pulnix (San Jose, CA) to develop machine-
image processors, and machine-vision soft ware. Unfortu- vision products.
nately there are no standard means to evaluate or program The European counterpart of the AIA, the European Ma-
these devices. chine Vision Association (EMVA; Frankfurt, Germany), is
In the world of computer graphics, Open GL and DirectX developing a new standard called GenICam ( www.genicam.
give developers a set of standard application programming org), the goal of which is “to provide a generic programming
interfaces (APIs) that provide access to the features of 3-D interface for all kinds of cameras.” Whether GigE, Camera
graphics-acceleration chips. Because most graphics pro- Link, or Fire Wire is used, the API would be always the same.
grammers and graphics-processor board vendors adhere to Currently, the GenICam standard consists of multiple
these standards, it is relatively easy to compare the perfor- modules including a GenApi to configure the camera, rec-
mance of one board to another. Things are not the same in ommended names and types for common features, a Trans-
the world of machine vision, where the lack of a common port Layer for grabbing images, and a DataStream to inter-
API has resulted in many greatly differing solutions. pret additional data that might be appended to the image.
Rather than wait for machine-vision suppliers, Sun Micro- The first version of the standard contains the GenApi mod-systems (Palo Alto, CA) and the DARPA sponsored Vector ule only, but others will follow.
Signal Image Processing Library ( www.vsipl.org) indepen- From a technical perspective, the question is: how will dently offer extensible imaging and video APIs designed to these standards incorporate any type of high-level program-fit between higher-level programming interfaces and system ming API? While it is important to develop standard pro-hardware. Supporting optional accelerators and frame buf- gramming interfaces that specify how to configure cameras fers, developers can use these APIs to add image processing and transport image data, the advent of smart cameras that functions and proprietary image-processing algorithms by incorporate frame-grabber and image-processing capabili-replacing defined operations through the device layer. ties is posing a more important problem. Rather than con-
Yet for system integrators, no machine-vision system ven- centrate on simple camera APIs, it will be far more impor-dor currently supports these APIs. In fact, instead of endors- tant in the future to adhere to APIs that provide developers ing a standard, many of those in machine vision seem bent with unified programming models. on developing simpler APIs that support different types of cameras. For example, the Automated Imaging Association CONARD HOLTON is editor in chief of Vision Systems Design; e-mail: (AIA; Ann Arbor, MI) has been working for the past two cholton@pennwell.com.
References:
Archives