Image analysis – Image transformation or preprocessing – Changing the image coordinates
Reexamination Certificate
1997-03-13
2001-03-27
Boudreau, Leo H. (Department: 2721)
Image analysis
Image transformation or preprocessing
Changing the image coordinates
C382S232000, C382S266000
Reexamination Certificate
active
06208768
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to the field of processing video information, and, more particularly, the present invention relates to an improved device and method of processing video display signals.
2. Description of the Related Art
The encoding of video signals in accordance with the image encoding standards H.261, H.263, MPEG1 as well as MPEG2 is often based on a block-oriented discrete cosine transformation (DCT). These block-oriented encoding methods, however, are not suitable for image encoding methods that are no longer based on rectangular blocks. For example, subjects from an image may be segmented and the segments of the image are encoded. These methods are referred to as region-based or subject-based image encoding methods. A segmenting of digital images thereby takes place according to the subjects occurring in the scene. A separate encoding of these segmented subjects is implemented instead of encoding the image blocks as in block-based image encoding methods. The encoding thereby usually is accomplished by modeling the segmented subjects and subsequent transmission of the modeling parameters for these segmented subjects.
After transmitting the image information from a transmitter to a receiver, the individual subjects of the image are in turn reconstructed in the receiver on the basis of the transmitted modeling parameters.
One possibility for modeling the subjects is a series development of the image function according to a set of suitably selected basic functions. The modeling parameters then correspond to the development coefficients of these image functions. Such a modeling of the image is the basis for transformation encoding. When individual, arbitrarily bounded subjects of the image are to be encoded, a transformation for segments with arbitrary, usually not convex bounds is required. Two basic approaches have previously existed for such a transformation.
In the method described in M. Gilge, T. Engelhardt and R. Mehlan, Coding of Arbitrarily Shaped Image Segments Based on a Generalized Orthogonal Transform, Signal Processing: Image Communication 1, 00. 153-180, October 1989, a given image segment is first embedded in a circumscribing rectangle having the smallest possible dimensions. A discrete cosine transformation (DCT) that is completely specified by the basic functions of the transformation can be recited for this rectangle. In order to match this transformation to the segment shape, the basic functions defined for the rectangle are successively orthogonalized with respect to the shape of the segment. The resulting orthogonal, shape-dependent basic functions then form the segment-matched transformation that is desired.
One disadvantage of this approach is that there is a tremendous demand for calculating capacity and memory for implementing this method. Further, with this known method no reliable statements can be made about the resultant transformation for data compression, since the transformation is essentially dependent on the orthogonalization sequence and, thus, on the specific implementation.
T. Sikora and Bela Makai, Shape-adaptive DCT for generic coding of video, IEEE Trans. Circuits and Systems for Video Technology 5, pp. 59-62, February 1995 describes a method wherein the given image segment is separately transformed according to rows and columns. To that end, all rows of the image segment are first aligned on the left and successively subjected to a one-dimensional horizontal transformation whose transformation length respectively corresponds to the number of picture elements in the corresponding row. The resultant coefficients are subsequently transformed a second time in vertical direction.
This method has the disadvantage that correlations of the brightness values for the picture elements (similarities of the picture elements) cannot be fully exploited due to resorting of the picture elements.
For improving this method T. Sikora, S. Bauer and Bela Makai, Efficiency of Shape-Adaptive 2-D Transforms For Coding of Arbitrary Shaped Image Segments, IEEE Trans. Circuits and Systems for Video Technology 5, pp. 254-258, June 1995 [3] describes a method wherein a transformation for convex image segment shapes adapted to a simple image model is implemented. However, only image segment shapes that exhibit no interruptions (holes) upon traversal of rows or columns are allowed.
In the above-described, known methods due to the variable transformation length, standard transformation methods or, respectively, standard modules can no longer be utilized for implementation of the transformation which is undesirable.
Standard image transformation methods are known, for example, from R. J. Clarke, Transform Coding of Images, Academic Press, London, pp. 72-134, 1985. Because standard image transformation methods can no longer be applied, substantial costs arise for encoding units that work according to the known, above-described methods.
The present invention specifies a method for processing picture elements of an image segment with an arbitrary shape that exhibits image segments of a target image segment shape. Standard image transformation methods and encoding devices are compatible and can continue to be utilized for implementation thereof. Other objects and advantages of the present invention will be apparent form the following summary and detailed description of the preferred embodiments when viewed in light of the drawings.
SUMMARY OF THE INVENTION
In the present invention, picture elements of an image segment to which brightness values are allocated are subjected to conformal imaging. One goal of this imaging is that at least the brightness values of the picture elements that are located at an edge of the image segment are imaged onto picture elements of an edge for a target image segment. After the brightness values have been subjected to this imaging, the brightness values are interpolated arbitrarily.
It is thus possible to also implement a subject-based image encoding with block-based image encoding methods without greater cost. Considerable cost savings are realized as compared to entirely new development of subject-based image encoders.
The present invention advantageously improves transformation results, not only for the brightness values of the edge picture elements of an image segment, but also for the brightness values of other picture elements of the image segment that are imaged. These are advantageously transformed into a region located between the area of the picture element that was allocated to the respective brightness value before imaging and the edge of the target image segment.
Additionally, it is desirable in the special case where the target image segment exhibits a rectangular shape, that the imaging of the brightness values takes place along a straight-line direction through the target segment in the straight-line direction for which the respective picture element whose allocated brightness value is imaged lies. This corresponds to that case where a block-based image encoding is to be subsequently applied to the picture elements of the target image segment. This simply corresponds to a “shift” of the brightness value along the corresponding straight line. What is especially advantageous about this aspect of the inventive method is the simplicity and, thus, the speed of implementing the invention.
Further simplification can be achieved in that conformal imaging ensues in such a way that the shift of the brightness values along the straight line does not take place only for the respective edge pixel. Rather, this occurs for all picture elements of the image segment, respectively proportional to an imaging factor that, for example, derives from the ratio of the size of the image segment to the size of the target image segment.
It is also advantageous that interpolation between the brightness values after the imaging takes place with an extremely simple and, thus, quickly implemented interpolation. This may be, for example, by a linear interpolation of the b
Kaup Andre
Pandel Juergen
Boudreau Leo H.
Desire Gregory
Schiff & Hardin & Waite
Siemens Aktiengesellschaft
LandOfFree
Method for processing picture elements of an image segment... 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 for processing picture elements of an image segment..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for processing picture elements of an image segment... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2441728