Kodak has been refining production scanners for over a decade to deliver the "Perfect Page," an optimized image capture process. In all that time, the fundamental requirements
driving design haven't changed.

Requirement #1: Make scanning as simple as possible. It's a means to an end; namely, supplying an imaging application with images. If we could, we would put images online by waving a magic wand. A Perfect Page-enabled Kodak scanner is the next best thing.

Requirement #3: The best-possible image quality is a must, because a bad image can be worse than no image at all. Imagine an e-mail system that scrambled text at random. How happy would you be if you had to make guesses or ask people to find and resend their messages? It would be like dropping the baton in a relay race. Perfect Page virtually guaran-tees first-pass image quality that you can count on, scan after scan after scan.

As you'll see from the discussion that follows, the beauty of Perfect Page Scanning is that it supports all three fundamental requirements. It makes image capture as simple, robust, and high-quality as it can possibly be.

In a Perfect Page world, production scanning takes place at high speed, without interruption. A scanner operator just loads and unloads paper. Meanwhile, the scanner converts all the visible information into clear, legible digital images that resemble the original document as exactly as possible, without operator guesswork. Downstream, these "Perfect Pages" are immediately readable by recognition programs and people alike. Labor-intensive document preparation and image clean-up after scanning are essentially eliminated.

The scenario gets worse. Intervention by people or computers before or after scanning adds time and expense to the process, delaying delivery of usable images to those who need to retrieve them. Throughput rates tumble because you may be forced to presort documents, make QC/QA adjustments to scanned images or physically rescan and reinsert images into a batch.

Also there's the time and expense of training (and retaining) scanner operators in the fine art of scanner control adjustments, which affect both content and quality. With estimates of labor's contribution to the total cost of scanning running as high as 76%, the fewer manual steps, the better.

Back to the Perfect Page capability designed into Kodak production scanners. It enables the scanners to optimize image quality while automating more of the work. This reduces labor and increases throughput to the host application.

With Perfect Page, what was on the page is what appears in the image. Just like a good first leg in the relay race, optimal scanner throughput and image quality contribute to winning performance throughout the rest of the imaging application.

FOR HIGH PERFORMANCE IMAGE CAPTURE, YOU PUSH THE BUTTON. PAGE PERFECT DOES THE REST.

To appreciate how Perfect Page streamlines and optimizes the document imaging chain, consider conventional photography. For best results, you use a good camera with a sharp lens, a precise shutter, and accurate exposure control. You load the camera with film that delivers consistently good quality – Kodak Gold Max Film, for example – and take your picture.

Before you can frame an 8-by-10 glossy and admire your work, the film must be processed and a print made. Any substandard component or poorly executed step in this series of events puts the final outcome at risk. Now look at the document imaging chain. Depending on a scanner's capability, you may have to sort documents into batches of similar contrast and color and scan them with varying scanner settings. Image capture takes place inside the scanner, where a lens system, lighting, and an electronic sensor take the place of camera, flash, and film. To make images ready for the host imaging system, they must pass through digital processing steps and human QC/QA, if required.

Just as in the relay race, each leg of the imaging race is dependent on all of the previous legs. For example, raw image capture that's less than optimum compromises the effectiveness of any subsequent digital processing.

Kodak has been continuously upgrading illumination, lens design, sensor performance, and processing hardware and firmware since the introduction of its first production scanners in the early 1990s. Guided by real-world experience and the voice of the customer, a steady series of innovations have improved every part of the document imaging chain. The result is Perfect Page Scanning.

" ...Kodak puts a lot of R&D into making an image that's as clear as possible. They've taken core technology that exists within the company and are building image manipulation into the scanner at no extra charge. Kodak is raising the bar with Perfect Page."

- Susan Moyse InfoTrends Research Group Inc.

Custom Optics

As noted above, Kodak custom-designs its own optics. Lenses are tuned for the spectrum of the scanner bulbs to provide superior light transmission and ensure that the scanner can be operated at rated speeds. Optical formulas are carefully specified to provide distortion-free, flat focus across the entire width of the paper path. Combined with calibrated illumination, the lens delivers consistently even contrast and sharpness from edge to edge, resulting in higher achievable resolutions.

Let's recap the Perfect Page story so far. As in the camera/film analogy, the all-important picture-taking step has taken place with accurate focusing and proper exposure. If it's wrong here, image processing won't save it later. Fortunately, the scanner has captured a great raw image using superior components, an optimized lens and a high-performance CCD array calibrated for the actual illumination. Now, just like the exposed image on the 35mm film, the raw image requires some processing.

PROCESSING IN COLOR
To optimize your color images for quality and file size, Kodak scanners use the high-speed computing power from the image processing engine. Through the use of specialized integrated circuits, built into Perfect Page-enabled Kodak scanners, you get high-quality color images that are quickly and easily deskewed and autocropped during scanning.

PROCESSING IN GRAYSCALE
Kodak scanners create optimum quality bitonal images of scanned documents by working with the grayscale representation of source documents. Grayscale processing makes better use of the raw image data, manipulating the image prior to converting it to bitonal. The image processing engine, which resides on a circuit board built into Perfect Page-enabled Kodak scanners, provides the speed and analysis needed to work with the grayscale information.

Without Perfect Page Scanning, image processing is often handled outside the scanner, via software on a PC.

With some external image manipulation software applications, the image processing chain starts with a bitonal, single-bit data stream. Deskewing bitonal images produces jagged lines and mis-shaped characters, reducing image clarity and OCR performance.

GETTING THE EDGE – ON ALL FOUR SIDES OF THE PAGE.

The first step in Perfect Page image processing is to determine where the document begins and ends. An algorithm called Contour Tracing searches the entire scanned image for outer boundaries. It isn't fooled by torn edges, successive skewed images, or cases in which the trailing edge of one document is close to the leading edge of the next. No part of the page is cut off as with ordinary image processing systems. Contour Tracing is integral to the Perfect Page, paving the way for highly accurate deskewing, autocropping, and/or border reduction.

PUTTING THE IMAGE ON THE STRAIGHT AND NARROW

The Perfect Page image processor "knows" the edges and up/down orientation of the page, thanks to Contour Tracing technology, and has the 8-bit data stream to work with.
In a step called grayscale deskew, it corrects any page rotation introduced during transport without creating jagged edges and staircasing artifacts.

PAINTING THE PERFECT PAGE, PIXEL BY CAREFUL PIXEL

The image still needs to be converted from grayscale to bitonal for the host imaging system. This process is called thresholding, and there's more than one way to do it Perfect Page image processing uses a computationally intensive approach called Adaptive Threshold Processing (ATP). This looks at the 48 neighboring pixels around each individual pixel in order to find the closest boundary transition between black and white.
Other processes look at just a few surrounding pixels. Which do you think would make it easier to identify a friend from a photo -- to look at the tip of his or her nose, the whole nose, or the eyebrows, eyes, nose, ears, mouth, and chin? ATP does the latter. The benefits are visible in cleaner, better-defined characters.
Depending on the model and applications, Perfect Page-enabled scanners can perform other image manipulations along with deskew. These may include autocrop, border reduction, error diffusion, and dithering. All occur faster than with ordinary scanning systems because they take place on-the-fly within the scanner, rather than in software on a PC.

Clearly, if you want to optimize quality and the overall scanning process, a Perfect Page-enabled scanner is the ultimate in color and black-and-white scanning. And there's no extra charge for the extra performance. Perfect Page is standard on the latest Kodak Scanners rated from 170 up to 640 images per minute. It's exactly what you need for color and black-and-white scanning that's simply perfect.

- Pete Rudak, Chief Technical Officer
Document Imaging, Eastman Kodak Company