Image analysis – Image enhancement or restoration – Edge or contour enhancement
Reexamination Certificate
1999-03-01
2002-04-16
Lee, Thomas D. (Department: 2724)
Image analysis
Image enhancement or restoration
Edge or contour enhancement
C382S275000, C382S254000, C358S447000, C358S463000
Reexamination Certificate
active
06373992
ABSTRACT:
BACKGROUND OF THE ART
This invention relates to a method and an apparatus for image processing for noise suppression and sharpness enhancement of digital images, or specifically to a method and an apparatus by which noise such as graininess in digital images can be suppressed while enhancing their sharpness.
In the present invention, the noise in digital image is called “grain or graininess” in view of the appropriateness of the expression if the image reproduced in photographs, printed documents and so forth is discussed and the noise region of a digital image being processed which consists of a noise component and which is to be separated from the edge region which consists of an edge (contour) component is called “a grainy or flat region” in view of the appropriateness of the expression if the image reproduced in photographs, printed documents and so forth is discussed.
In digital imaging technology which obtains pictures such as photographs with an image input scanner and which outputs digital images with an image output printer, considerable deterioration occurs in the sharpness of the output image due to the scanner and the printer. As a corrective procedure, sharpness enhancement is conventionally performed by means of a Laplacian filter or an unsharp masking (USM). However, sharpening the image causes the side effect of increasing noise such as graininess and, hence, grainy pictures can be subjected to only moderate sharpness enhancement within a range where graininess deterioration is tolerated; as a result, it has been difficult to obtain image quality better than that of the original grainy picture.
Several methods have been proposed to process digital images such as to remove noisy graininess and enhance their sharpness. Removal of graininess involves an averaging or blurring technique but the blurred grainy pattern is not pleasing to the eye or fine structures of the object may be erased in an unnatural way. For these and other reasons, the conventional techniques for removing graininess are not suitable for application to high-quality pictures such as photographs.
Pictures such as those in photographs, printed documents, or on television's screens and from various kinds of copiers suffer diverse deterioration problems, i.e., sharpness deterioration due to optics such as a camera, graininess and sharpness deterioration inherent in photographic materials, or noise and sharpness deterioration that is added when the original picture such as a photograph or a printed document is digitized with an image input device. In order to deal with these difficulties, various methods have been proposed to process images such as to reduce noise and enhance their sharpness. Smoothing and coring are two common methods employed in the conventional image processing technology for removing graininess, whereas sharpness enhancement is implemented by unsharp masking (USM) or processing with a Laplacian or a high-pass filter. However, if graininess is suppressed by these conventional methods of graininess removal, artifacts that cause unnatural and strange impressions will occur or fine structures of the image that should inherently be kept intact will be suppressed along with the graininess. See, for example, Japanese Domestic Announcement (koho) Nos. Sho 57-500311 and 57-500354, as well as P. G. Powell and B. E. Bayer, “A Method for the Digital Enhancement of Unsharp, Grainy Photographic Images” in the Proceedings of the International Conference on Electronic Image Processing, Jul. 26-28, 1982, pp. 179-183. According to the method proposed by Powell and Bayer in these references, suppression of graininess is accomplished by smoothing (with a low-pass filter) and sharpness enhancement is performed with an unsharp masking (high-pass filter). In the smoothing process, signal values for n×n pixels are multiplied by Gaussian or other type of weights such that the signals are smoothed to suppress graininess. In the sharpness enhancement process, picture signals for m×m pixels are first used to determine differential values by calculation from the central pixel towards the surrounding pixels and if any differential value is smaller than a preset threshold, the pixel of interest is regarded as representing graininess or noise and removed by coring and the remaining differential values which are greater than the threshold are summed up, multiplied by a constant more than 1.0 and added to the previously smoothed signals, thereby achieving sharpness enhancement.
In this process, the density contrast of grainy patterns decreases since they are blurred; on the other hand, blurred grainy patterns may become visually pronounced as they are mottles of randomly crowded grains that cause graininess (this phenomenon is commonly referred to as “mottling”) and they will present unpleasing graininess. In addition, a preset threshold is used as a criterion for distinguishing graininess from the picture (this is the coring process), so image signals of low contrast may occasionally be erroneously taken as graininess and suppressed or removed along with the latter or discontinuity may occur at the boundary between the removed image signal and the enhanced picture signal to produce an unnatural artifact in the output image. This drawback occurs very frequently in fine images such as those of lawn and carpets and in images that represent texture as in fabrics and the result is an artifact that is visually quite unnatural and hence undesirable.
In the above-described prior art method of processing images to suppress their graininess while enhancing their sharpness, unsharp masking is employed to enhance the sharpness whereas blurring or smoothing is effected to suppress the graininess, such that a graininess (noise) signal and a contour signal are separated from the original picture by signal level and the contour signal is subjected to sharpness enhancement whereas the smoothed region is suppressed in graininess and the smaller signal is regarded as representing graininess and processed accordingly; as a result, signals representing the fine details of the picture which are close to signal levels representing graininess, namely, image signals representing the texture of cloths, the hair on the head and the like, will be suppressed along with the graininess, yielding visually unpleasing images that contain artifacts from image processing In the conventional image processing technology where blurring or averaging is used as the method of suppressing graininess, a blurred grainy pattern is reduced in terms of density fluctuation; on the other hand, blurred grainy pattern spreads despite the small amount of density fluctuation and will be recognized as a visually unpleasing pattern, which stands out markedly in someone's face or skin as in portraits or in solid objects such as walls or sky.
In the prior art, a grainy (noisy) signal region and a contour region are separated from the original picture by signal level. Stated more specifically, the contour region and a flat region are discriminated on the basis of a signal indicating the difference between the original picture and a blurred image and the respective regions are processed with an unsharp masking, a Laplacian filter or other suitable means using different coefficients such that graininess is suppressed in the flat region whereas sharpness is enhanced in the contour region, thereby achieving graininess suppression without producing blurry edges. A problem with this technique is that discontinuity will occur at the boundary between the contour and grainy regions because the recognition and separation of these regions are performed indiscriminately with reference to a single threshold signal level.
It should also be mentioned that in the prior art which employs unsharp masking or a Laplacian filter for edge or sharpness enhancement, fringe (over shoot) such as Mach bands are most likely to occur along the contour or edges of the image, giving the visual impression of artificiality.
SUMMARY OF THE INVENTION
The present invention has been
Fuji Photo Film Co. , Ltd.
Lee Thomas D.
LandOfFree
Method and apparatus for image processing does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for image processing, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for image processing will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2843204