Image analysis – Image compression or coding – Adaptive coding
Reexamination Certificate
2001-04-09
2003-09-09
Ahmed, Samir (Department: 2623)
Image analysis
Image compression or coding
Adaptive coding
C382S172000, C358S426140, C348S409100
Reexamination Certificate
active
06618509
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention generally relates to coding digital video images, and more particularly to reducing loss of image information by automatically adjusting operating parameters utilized in the coding process.
2. Description of Background Art
Digital video systems are becoming increasingly popular, especially in business settings. An example application of a digital video system is a teleconferencing system. Despite their popularity, digital video systems can be extremely expensive in terms of storage and communication costs. The cost of storage and communication is driven by the massive quantity of digital image data which is generated by the system.
One way to reduce costs or improve performance is to reduce the quantity of digital data used to represent images. Various well known compression techniques have been utilized to reduce the quantity of data used to represent a digitized image. While image compression may reduce some of the costs associated with handling digital image data, the downside is that image quality may suffer.
A number of compression techniques conventionally involve linear transformation of the digital image, followed by quantization, and coding of transform coefficients. In this way, the quantized and coded signals may be compressed, transmitted, or stored, and subsequently decompressed using an inverse set of operations.
The Discrete Cosine Transform (DCT) has commonly been used for image compression and decompression. However, because such DCT-based image processing is computationally intensive, various methods have been devised to improve the performance of the transform process.
The DCT process involves computing a set of coefficients to represent the digital image. One approach used to reduce the time required to perform the transform process is to compute only a subset of the coefficients. The selection of the particular subset of coefficients to be computed is based on detected characteristics of the digital image. While yielding acceptable results, the prior art process of classifying a digital image according to its characteristics and then selecting a subset of coefficients has no mechanism to measure the quality of the transformed image. Furthermore, the selection criteria used to classify an image are fixed such that they cannot be easily adjusted to improve image quality.
Therefore, to improve the quality of compressed digital images what is needed is a coding system having self-adjusting selection criteria for selecting a transform function.
SUMMARY OF THE INVENTION
The invention monitors the quality of coded digital images, and based on the monitored quality of the images, updates operating parameters that are used in coding the images.
A set of predetermined coding functions is available in a video coding system to code a digitized video image. One of the coding functions is selected and applied to the input image. The selection of the coding function is made based upon measured characteristics of the input image and selection criteria which are applied to the measured characteristics. The image is then decoded and the quality of the decoded image is measured. The selection criteria are updated based on the measured quality of the decoded image, whereby for subsequent images coding functions are selected to produce images with a higher quality measure.
In another aspect of the invention, an historical record is made for the measured characteristics of the images processed by the system. The measured characteristics are correlated with the selected coding function. Periodically, the selection criteria are updated based on the historical record. The historical record provides a broad perspective upon which updating of the selection criteria is based.
The invention further selects one of a predetermined set of transform functions to code an image. An inverse transform function is selected, independent of the selection of the first transform function, whose application minimally covers the coefficients produced by application of the first transform function. The inverse transform function is then applied to the coefficients, the quality is measured, and the selection criteria are updated as described above. The updating of the selection criteria enables selection of a suitable transform function.
In still another aspect of the invention, the selection criteria include adjustable thresholds and comparisons of them to measured characteristics of the image to be coded. The measured characteristics are correlated to the selected inverse transform function in the historical record. The respective thresholds are then updated from the historical record of the measured characteristics.
REFERENCES:
patent: 5327502 (1994-07-01), Katata et al.
patent: 5559900 (1996-09-01), Jayant et al.
patent: 5719961 (1998-02-01), Normile et al.
patent: 5883979 (1999-03-01), Beretta et al.
patent: 6229917 (2001-05-01), Wu et al.
Performance of a real-time software-only H.261 codec on the Power Macintosh, H. Wu et al., SPIE vol. 2419 pp. 492-498.
Chu Ke-Chiang
Lu Jian
Tian Yu Tina
Wu Hsi-Jung
Ahmed Samir
Apple Computer Inc.
LandOfFree
Digital image coding system having self-adjusting selection... 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 coding system having self-adjusting selection..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Digital image coding system having self-adjusting selection... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3014619