How to select components for image processing

How to Select Components for Image Processing by Robert Gregory Senior Product Marketing Manager for Imaging Products, Data Translation, Inc.

Advances in bus architecture and processor speeds have made it possible to use PC-based imaging systems for applications unheard of just a few years ago.

Although technology is quickly changing, setting up such a system will be easier if you consider some basic guidelines.

Because the applications of image processing are so diverse, components can vary from system to system.

However, components usually fall into several set categories: - Input Source and Optics - This can be a camera, microscope, scanning electron microscope, or CCD array.

These devices are used to take a picture, or "image", of the object being inspected.

Depending on the application, cameras can be standard monochrome (also known as RS-170), composite colour (or Y/C), RGB colour, non-standard monochrome (variable scan), line-scan, or custom CCD arrays (used for X-ray).

Cameras can be built into a microscope or mounted separately.

- Lighting - Lighting is needed to illuminate the object or specimen so that the input device can acquire the best possible image.

Fixtures and bulbs are available in a wide variety of shapes, sizes, and intensities, but common bulb types are fluorescent, LEDûs (Light Emitting Diodes), and high-intensity.

- X-Y Table - This component automates the process of acquiring images of multiple samples.

The table moves a predetermined distance after each image is acquired to properly position the next object or specimen in relation to the camera.

- Frame Grabber - This component is a circuit board that takes an analogue image from the camera and converts it to digital information for use by the PC.

Frame grabbers are installed inside the PC chassis, and come in various configurations to support different camera types (for example, monochrome, colour, or line scan) and computer bus platforms (for example, PCI, ISA, and VME).

For PC-based image processing, the PCI bus is the most common.

- PC Platform - The computer that is the heart of an imaging system.

For this application, the computer usually is equipped with a 133 MHz (or faster) Pentium or Pentium II processor, and Windows 95 or NT. The PC should have a minimum of 32 MB of RAM, especially if you will be processing highly detailed images.

To determine the memory needs for your application, it may help to decide how much memory each image will take up.

For an eight-bit monochrome image, simply multiply width x height to compute the memory needed.

For example, if the image size is 1K x 1K, then 1 MB of memory is required for each image.

If the resolution is more than eight bits, then the formula is width x height x 2.

- Imaging Software - The software allows the image to be processed, analysed, and stored.

Different types of software packages are available, ranging from easy-to-use packages with pre-defined tools, to SDKs (software development kits) that allow programmers to build custom imaging applications.

- Network Connection - Once the system has acquired an image, that image and the resulting data may need to be accessed by other users.

This is accomplished with a network connection that links computers together, usually within an organisation.

Careful planning will ensure that your system meets your needs.

The following is a checklist of the issues you should consider.

What are your goals- - Decide what you want to accomplish with your system.

Imaging operations fall into several categories including measurement, counting, particle classification, blob analysis, etc.

Also consider how many samples you want to process, now and in the future.

Select the right components - A PC-based imaging system is only as strong as its weakest link.

Any shortcuts made in component selection, especially in the optics and imaging path, can greatly reduce the effectiveness of a system.

The basics to remember in choosing the components of the image path are: - Cameras - Typically, camera selection is directly tied to the application requirements.

For example, a monochrome camera is usually used for general applications while a colour camera would be needed for colour applications.

In addition, the resolution should be high enough to capture needed information.

For the highest accuracy, a digital camera may be the best option.

- Optics and Lighting - These crucial considerations are often overlooked.

When poor optics or lighting is used, even the best system will not perform as well as a less capable system with good optics and adequate lighting.

With the right optics, you will get the best and largest usable image, and with correct lighting, you will illuminate the key features being examined.

In choosing lighting for your system, first consider the other lighting in the room. Certain types of room lighting may interfere with the capture rate in the camera, resulting in artefacts on the image (often referred to as a herringbone pattern).

Use of high-frequency fluorescent fixtures will avert this problem.

Choosing lighting for microscopy applications poses a specific set of challenges. Pay particular attention to the illumination underneath the microscope.

If the lighting is not built in, select lighting that is completely even, with a colour spectrum that best illuminates the objects being examined.

For applications involving live samples, avoid fixtures that generate damaging heat.

- Frame Grabbers - Although the frame grabber may be one of the smallest parts of a machine vision system, it is one of the most important, since it too is in the image path.

As mentioned earlier, the frame grabber converts the analogue image data from the camera into digital data for use within the PC.

Ideally, the frame grabber should do this with as little alteration to the image data as possible.

Choosing the wrong frame grabber can result in errors being introduced into this image data.

- Software - Finally, the right software will help you reap the benefits of your careful hardware choices.

Options range from complete, easy-to-use Windows-based application packages with graphical user interfaces, to basic packages with add-on modules for specific functions, to software development kits that allow programmers to create their own custom applications.

Many manufacturers offer demo versions of their software products, so by all means try before you buy, keeping your specific application needs in mind.

Minimise possible problems - The human eye and brain are elaborate and versatile systems, capable of identifying objects in a wide variety of conditions.

For example, we are able to identify familiar people even when they are wearing different clothes, and recognise familiar landmarks when driving on a foggy day.

A PC-based imaging system is not as versatile; it can only perform what it has been programmed to perform.

Knowing what the system can and cannot "see" are important points to keep in mind to obtain the results you want, and reduce errors and incorrect measurements.

Common variables include: - Changes in object's colour - Changes in surrounding lighting - Changes in camera focus or position - Improperly mounted camera - Environmental vibration A vibration-free environment with all extraneous light removed will eliminate many common problems.

Store and access the data - For many systems, obtaining an accurate image is only the first step. The capability to store the data and easily access it later by others is just as important.

Careful planning will be needed at this stage to produce a smooth integration of the imaging system into a computer network.

Common questions that must be addressed are: the types of signals required, the kind of network currently used or required, and the file formats that will be transferred. Another consideration is whether the software youûve selected will support the networking function.

A PC-based imaging system is only as good as its weakest link, and only as accurate as the information it receives.

Spending some time and effort in performing the proper set-up, and paying attention to the details, will result in an accurate and resilient system.