Image analysis – Image compression or coding – Transform coding
Reexamination Certificate
1998-03-06
2001-06-12
Grant, II, Jerome (Department: 2724)
Image analysis
Image compression or coding
Transform coding
Reexamination Certificate
active
06246801
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to image encoding and decoding techniques, and more particularly, to a method and apparatus for generating a partial image from a larger compressed image (or a plurality of individual images).
BACKGROUND OF THE INVENTION
Due to the limited bandwidth of transmission channels, there are a limited number of bits available for encoding image information, such as image information generated by a camera for transmission to one or more remote users. Thus, there are many image encoding techniques available which encode the image information with as few bits as possible using compression techniques, while still maintaining the quality and intelligibility that are required for a given application.
Remote cameras, such as those used for security applications, traffic monitoring or daycare monitoring, are typically panned by physically moving the camera. In addition to the possibility of a mechanical failure, the utility of such remote cameras is limited in that only one user can control the camera at a time. For multi-user applications, however, such limited user control of the camera view is not practical. A number of software techniques have been developed for permitting a number of users to view selected portions of a larger image (or a composite image generated from a plurality of individual images).
Permitting multiple selected views of a larger image, however, becomes more difficult if the larger image is compressed. Specifically, since image data following image compression is of variable length, pixel boundaries are not readily detectable in a compressed image. In addition, since many encoding techniques exhibit intra-frame pixel dependencies, such as encoding the difference values for adjacent DC coefficients under the JPEG standard, the pixel values must be modified when generating a selected portion of a larger image, to reflect the reordering of the subset of pixels in the selected image view.
Typically, when generating a selected portion of a larger compressed image, the larger image must be decompressed into the pixel domain, before the pixel values are reordered and assembled to create each of the selected image views. Thereafter, each of the selected image views is compressed to form the final images transmitted to each user. The more popular image compression techniques, such as JPEG and MPEG, typically perform three steps to generate a compressed image, namely, (i) transformation, such as a discrete cosine transform (DCT); (ii) quantization; and (iii) run-length encoding (RLE). Likewise, to decompress images using these same image compression techniques, the inverse of the compression steps are performed by the receiver on the compressed image, namely, (i) run-length decode; (ii) dequantization; and (iii) inverse discrete cosine transform (IDCT).
Thus, to create N selected image views from a larger compressed image, conventional techniques require one image decompression, N pixel reorderings, and N compressions.
SUMMARY OF THE INVENTION
Generally, a selected image view generator for generating a selected portion of a larger compressed image is disclosed. The larger compressed image includes a plurality of macroblocks of image data, encoded using an intraframe encoding technique. The macroblocks do not need to be encoded independently. A portion of the DC coefficient values, however, are recomputed when generating the selected image view from the overall image, to reflect the reordering of the pixels in the selected image view (that results in a portion of the pixels being adjacent to different pixels than they were in a subimage).
According to a further aspect of the invention, a selected image view is generated from a larger compressed image having a plurality of macroblocks of image data. Each of the macroblocks contains one or more blocks of data, with each block having a DC coefficient value. Initially, the blocks from the larger compressed image that are included in the selected image view are identified, and then the DC coefficient values are recomputed for each block in the selected image view that is encoded relative to a different block than it was encoded relative to in the larger compressed image. Finally, the identified blocks are assembled to form the selected image view.
According to a further aspect of the invention, multiple users can simultaneously control a selected view received from an image source. Remote cameras can be panned electronically, rather than physically. The overall image may include one or more static or real-time images. The selected image from a larger overall image may be used, for example, with a 360° panning camera, to permit each user to select a desired real-time or static view. For example, for a tennis match, different users can watch different players from the same video feed.
REFERENCES:
patent: 5262855 (1993-11-01), Alattar et al.
patent: 5442400 (1995-08-01), Sun et al.
patent: 5481297 (1996-01-01), Cash et al.
patent: 5533138 (1996-07-01), Kim et al.
patent: 5623308 (1997-04-01), Civanlar et al.
patent: 5657092 (1997-08-01), Kim
patent: 5666487 (1997-09-01), Goodman et al.
patent: 5878168 (1999-03-01), Kondo et al.
patent: 5912742 (1999-06-01), Matoba et al.
Boyce Jill MacDonald
Boyer David Gray
Grant II Jerome
Lucent Technologies - Inc.
Ryan & Mason & Lewis, LLP
LandOfFree
Method and apparatus for generating selected image views... 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 generating selected image views..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for generating selected image views... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2521302