XEROX INNOVATION AT WORK: AUTOMATIC IMAGE ENHANCEMENT

Send to friend

Latest in Xerox digital colour image technology now added to Workflow software

Xerox scientists have announced the latest in Automatic Image Enhancement (AIE) technology. This newest version of AIE can correct the shortcomings found in images clipped from a Web page or compressed too tightly for faster e-mail. It also works on specific problem spots and incorporates a new algorithm that can automatically determine in advance whether or not an image will be improved by applying enhancements to those areas.

While software exists to manually fix digital image problems, it requires time and a skilled craftsman. In the late 1990s, Xerox, which boasts some of the worlds top scientists in the fields of digital imaging and colour science, took up the challenge of developing a system that could automatically correct imperfect images without damaging good ones. The solution developed by its researchers is Automatic Image Enhancement technology, or AIE, which has automated colour correction in Xerox systems. 

The basic AIE algorithms significantly enhance pictures brightening underexposed images, sharpening blurred prints, or burning the haze off holiday images to let the bright colours shine through. When originally introduced, AIE corrected the most common amateur photography mistakes.

AIE technology can be applied in the print driver on the users computer, as a pre-press service, or within printer software itself. In all cases, it analyses the image and, if needed, develops a custom prescription of image processing elements and applies it to the image.

Found only in Xerox digital systems, ranging from desktop colour laser printers to 54-inch-wide colour printers/plotters, AIE has just been added to the latest version of Xerox FreeFlow Process Manager software, which is part of Xeroxs FreeFlow Digital Workflow Collection for commercial print providers. In this latest version, AIE not only fixes contrast in sections of photos but also includes features relevant to commercial printers who want to be able to fine-tune electronic images in portable document format workflows.  Specifically, the new version includes:

  • Local Contrast Enhancement.  Until now, AIE algorithms were applied to the entire image, making it lighter or darker, or adding or reducing contrast.  But a common problem with photographs is poor lighting, where only a section of an image is too dark or too light because of back lighting, combined indoor/outdoor scenes, strong shadows or flash photography. For cases where there is a problem in only a portion of the photo, scientists at the Xerox Research Centre Europe, the companys research and development facility in Grenoble, France, have developed a patented solution based on a combination of heuristic and machine learning approaches. The new technology is very fast and can automatically optimise an images exposure region-by-region. It employs a decision mechanism to anticipate the effect of the enhancement and will only make the change when it is sure that the effect will be positive. The success of the algorithm has been validated by extensive user studies.
  • Expert Interface.  AIE was originally developed as a quick, simple one-button solution to improve print quality.  For commercial printers who want more control over the appearance of their images, Xerox scientists have added an expert interface that allows print professionals to fine-tune contrast, sharpness, and the lightness and darkness of the print.
  • PDF Functionality.  In prior implementations, AIE worked on images, which then were inserted in a PDF file for printing.  The new version makes AIE convenient for print shop professionals because it manipulates images within the PDF file.

The important new applications for digital colour printing, such as one-to-one marketing and short-run printing, demand that we reduce the time and skills required to create a document, says Reiner Eschbach, a Xerox research fellow.  Our Automatic Image Enhancement technology is part of Xeroxs answer.

Comments (0)

Add a Comment

This thread has been closed from taking new comments.