Automatic Photographic Image Adjustment Technology

Along with the spread of digital cameras, the range of ways in which digital images are used has also expanded to include prints of personal snapshots and photo albums, business documents, product catalogs, and photography books. Ideally, users would be able to take photos and reproduce them exactly as they had envisioned, according to their intended purpose. However, sometimes, due to reasons such as lighting conditions, it can be difficult to take photos that appear as intended. Furthermore, when photos are printed, memory colors (certain colors that leave a strong impression in the minds of viewers, such as the colors of people's skin) may not reproduced according to users' expectations. In these situations, the images may be corrected, but this requires technical knowledge and tools, and the correction process is time-consuming.
To resolve these problems, Fuji Xerox has developed automatic photographic image adjustment technology that makes it possible to reproduce images photographed with a digital camera according to the user's intent.

This automatic photographic image adjustment technology has two functions. The first function is used to automatically correct undesirable characteristics of an image photographed with a digital camera (the original image), and the second function is used to adjust memory colors to more preferable colors. The processing for these two functions is performed in succession as follows. First, the luminosity, color balance (white balance), contrast, and chroma of the original image are analyzed, and undesirable characteristics are automatically corrected. This processing also neutralizes the color cast that sometimes occurs in photos due to reasons such as the color of the light source or differences in illuminance. Next, the adjustment processing for memory colors is performed. For this processing, target values for memory colors are set automatically in order to prevent the image as a whole from having an unnatural appearance. It is therefore possible to adjust only the predefined memory colors (such as the colors of people's skin, the sky, trees, and grass) in applicable areas of the image to be more preferable colors without affecting the colors of other areas of the image.
Fig. 1 shows an example of skin tone adjustment. First, the processing to improve undesirable characteristics is performed on the original image (a) to create the image after improving undesirable characteristics (b). Then, the regions showing human skin are isolated (c), and from those regions, a representative color (a color characteristic of the skin in those regions) is calculated. Next, in order to ensure that the skin tone in the image is reproduced as a more preferable color when the image is printed, the color balance of the image as a whole is used to calculate appropriate target values for the skin tone, and adjustment processing is performed only on the regions showing human skin, creating the adjusted image (d).

Fig. 1: A flowchart of the automatic photographic image adjustment process

In addition, we have also developed technology to correct the visual perception of images. This technology, which is able to automatically adjust images according to the look desired by the user, is available as a feature on some of our production printers. After the processing to improve undesirable characteristics of the original image and the processing to adjust memory colors have been performed on an image, this processing to correct visual perception can be used to automatically perform chroma contrast correction and spatial frequency correction on the image, giving it the desired look. The technology can be used to automatically correct an image in various ways, such as to make it look more vibrant (Fig. 2: Emphasizing image vibrancy) or to give it a more detailed look (Fig. 2: Emphasizing image detail).

Fig. 2: Examples of correcting the visual perception of images (emphasizing image vibrancy and emphasizing image detail)