Computer graphics processing and selective visual display system – Computer graphics processing – Graph generating
Reexamination Certificate
1999-07-06
2001-08-21
Zimmerman, Mark (Department: 2671)
Computer graphics processing and selective visual display system
Computer graphics processing
Graph generating
C345S442000, C345S440000, C345S440000, C345S182000, C345S182000
Reexamination Certificate
active
06278463
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates generally to digital image processing systems for joining together digital images to produce a composite image. Embodiments of the present invention provide methods and apparatus for processing digital image data to produce digital image data representing such a composite image.
BACKGROUND OF THE INVENTION
Systems for joining two or more digital images to produce a composite digital image are known. Joining of images in this way may be required for a variety of reasons, but it is usually performed to obtain a larger field of view in the composite image. Generally, each pair of adjacent images to be joined to produce the composite image will have a region of overlapping image content. Such images can be acquired, for example, by separate cameras with overlapping fields of view, or, for stationary scenes, by shifting the field of view of a single camera between image acquisitions. However acquired, there will almost always be differences in image quality, such as color or brightness differences, between the individual component images, and these differences tend to accentuate the joins in the composite image. Prior known systems attempt to reduce the visibility of the join between two component images by processing image data from the overlap region such that pixels from both of the component images are combined in some way to produce the overlap region in the composite image. For example, each pixel in the overlap region of the composite image might be produced by averaging the corresponding pixels in the two component images. Alternatively, a more complex filtering process might be performed across the overlap region to generate the composite image pixels from the component image pixels. One common technique is to implement a gradual transition between the two component images by gradually ramping down the luminance level of one image signal across the overlap region and gradually ramping up the luminance level of the other signal across this region, the two image signals then being summed to produce the composite image signal. This produces a gradual fade from one image to the other across a straight-sided band corresponding to the overlap region.
While these known systems can reduce the visibility of the join to some extent, the resulting band in the composite image across which the processing has been performed can still be quite noticeable to the viewer. Moreover, the processing operations performed to obtain this band are relatively computationally intensive, involving floating point calculations, and this in turn limits processing speed. This can make implementation in real-time video applications more difficult, particularly in the case of high resolution video.
DISCLOSURE OF THE INVENTION
According to the present invention there is provided a method for processing digital image data representing first and second images with overlapping image content to produce digital image data representing a composite image spanning the overlap region, the method comprising processing image data which corresponds to the overlap region in the first and second images to produce image data representing the overlap region in the composite image such that the proportions of first image data and second image data contributing to the overlap region in the composite image vary along the length of the overlap region.
In embodiments of the present invention, therefore, data from the first and second images are combined to produce the overlap region in the composite image, but in proportions which vary along the length of the overlap region, i.e. in a direction perpendicular to the direction of increasing overlap. In the prior systems mentioned above, the relative contributions of first and second image data to the overlap region in the composite image are either uniform throughout the overlap region or vary only in a direction across the overlap region, i.e. perpendicular to its length. This lack of variation in the length direction means that there are straight-line boundaries between the different regions of the composite image down the whole length of the overlap region. Since the human psychovisual system is particularly sensitive to long straight lines, the join in the composite image is often still quite noticeable in these prior systems in spite of the relatively complex processing. In contrast, by varying the relative proportions of first and second image data along the length of the overlap region in embodiments of the present invention, the appearance of these straight-line boundaries can be reduced or eliminated, with the result that the join appears far less noticeable to the viewer. In other embodiments, results which are similar to or better than those obtained with prior systems can be achieved by a simpler processing operation, enabling equivalent or better image quality to be achieved with significantly improved processing speeds.
Preferred methods embodying the invention include defining a non-straight line extending along the length of the overlap region, and performing said processing by reference to said non-straight line such that the ratio of the proportions of first and second image data contributing to the overlap region varies for different sections of the overlap region along the length thereof. Various ways in which the processing can be performed by reference to said non-straight line will be described below, but the underlying idea is to generate pixels in the overlap region of the composite image in different ways depending on their location relative to this line to achieve the variation in said ratio. In particularly preferred methods, the ratio of the proportions of first and second image data contributing to the overlap region on one side of said line is 1:0, and the ratio of said proportions contributing to the overlap region on the other side of said line is 0:1. Thus, in the overlap region of the composite image, the image portion on one side of the line is made up solely of first image pixels, and the image portion on the other side of the line is made up solely of second image pixels. In a particularly simple case where the line divides the overlap region into discrete sections corresponding to image portions on alternate sides of the line along the length of the overlap region, the ratio of said proportions is preferably alternately 1:0 and 0:1 for successive sections along the overlap region. Again, therefore, portions of the overlap region in the composite image are made up solely of either first or second image pixels. These embodiments provide extremely simple methods for joining images since each composite image pixel is either a first or second image pixel, and averaging, filtering or other processing involving multiplication and summation of pixel values is not required. However, because the boundary between first and second image pixels along the length of the overlap region is a non-straight line, the join can be far less noticeable in the composite image. If the scale is small enough so that the eye cannot resolve the individual sections of first and second image pixels, then the apparent effect will be a merging of the first and second images across the join line, with the eye effectively performing an averaging process. Here, results which are as least as good as prior averaging systems can be achieved with a simple, high speed processing operation which does not involve floating point calculations. Alternatively, if the scale is such that the sections can be resolved by the eye, it may be desirable, depending on the requirements of a particular application, to generate composite image pixels in a band of pixels axially disposed about the join line by combining proportions of first and second image pixels in accordance with an averaging, filtering, or ramping process. While this would necessitate floating point calculations and thus increase processing complexity, very high quality results can be achieved since the resulting band of processed pixels has non straight-line boundaries which a
Chapman Sydney George
Collins Brian Michael
Herzberg Louis P.
International Business Machines - Corporation
Padmanabhan Mano
Scully Scott Murphy & Presser
Zimmerman Mark
LandOfFree
Digital 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 Digital image processing, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Digital image processing will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2549617