Image Processing

Document Scanner-based Image Enhancement

Manufacturers of today's production document scanners strive to deliver high quality images. One of the ways that they do it is by including built-in image enhancement hardware. Given different brand names by the respective vendors, Kodak's PerfectPage for example, circuit boards on the scanner take the original image and attempt to improve the contrast, clarity, and legibility of the image before it is passed to the host computer.

Hardware Image Processing Once an image has been created on the scanner, it is send to the "host" computer via a cable to a host interface. Some vendors offer scanner interfaces with built-in image processing. They receive the image, then apply additional image processing. Some of the hardware can even to rudimentary recognition of "patch codes" and bar codes in the short period of time before the image is saved to disk as a file.

Advanced Image Processing: VRS

More recently, hybrid hardware technology is on the market that combines hardware on the scanner with hardware in the host computer. Kofax's Virtual Rescan, or VRS is designed to maximize image quality over a wide range of input paper quality. It works with a card in the scanner that delivers the original grayscale image to the host computer, where the VRS-enabled hardware dynamically recalculates the thresholding needed to convert the image to black and white.

 

Software Image Processing

Hardware-based image process during scanning has only a very limited amount of time to work on the image before the next image arrives - on an average production scanner running 60 pages a minute - there is a just one second per image available. On faster scanners, even less time is available. To really enhance the image, the processing needs to be freed from the time constraint. Running images through software image processing modules after scanning is complete allows a lot more time to enhance the image through multiple passes.

Datacap scan software supports VRS and all hardware-based image processing. When more nuanced image processing is required, a separate Taskmaster task performs the honors. For maximum control, configure Taskmaster rules to sequence the image processing as required, perhaps first deskewing the image, then removing horizontal lines.

Grayscale Images

A black and white image has one piece of information per pixel. Any enhancement involves deciding whether that images should be on or off. It is always possible that the "enhancement" ends up making the wrong decision about a pixel, resulting in poorer quality, instead of better. There is not much information to work with in this environment, which is why alternative strategies are used in products like VRS.

Datacap uses grayscale images, where there are 256 possible values for each pixel, as the basis of image processing. If the scanner delivers a grayscale or color image, then the software has much more information about each pixel to work, with before it needs to make a decision. The Taskmaster Grayscale Conversion Task takes grayscale images and then, using Datacap algorithms, converts them from grayscale to black and white.

Similarly the Taskmaster Color Conversion Task takes color images and converts them to grayscale or black and white.