TH41508B - Adaptive weighting of the reference image in video decoding? - Google Patents

Abstract

DC60 (12/09/46) Adaptive reference image weighting in video decoding, decoders, video encoders, and corresponding methods for processing video signal data for image blocks, and special reference image indices for predicting image blocks, are revealed. The use of adaptive reference image weighting to enhance compression is exploited, where the decoder includes a reference image weighting factor unit to determine the weighting factor corresponding to the special reference image index. The encoder includes a reference image weighting determiner to determine the weighting factor corresponding to the special reference image index, and the decoding method includes receiving a reference image index with data corresponding to the image block, determining the weighting factor for each received reference image index, retrieving the reference image for each index, compensating for the motion of the retrieved reference image, and multiplying the motion-compensated reference image by the corresponding weighting factor to generate a weighted motion-compensated reference image. Adaptive reference image weighting in video decoding, decoders, video encoders. The corresponding methodology for processing video signal data for image blocks and special reference image indices for predicting image blocks is revealed. It utilizes adaptive weighting of the reference image to enhance compression, where the decoder includes a reference image weighting unit to determine the weighting factor corresponding to the special reference image indices. The encoder includes a reference image weighting determiner to determine the weighting factor corresponding to the special reference image indices, and the decoding method involves receiving a reference image index containing data corresponding to the image block, determining the weighting factor for each received reference image index, retrieving the reference image for each index, compensating for the movement of the retrieved reference image, and multiplying the motion-compensated reference image by the corresponding weighting factor to generate a weighted motion-compensated reference image.

Claims (17)

1. A video decoder is a device for processing video signal data for an image containing a number of image blocks. The decoder consists of a reference image weighting unit to receive at least one reference image index, one index of at least one reference image index for independent identification without using other indexes, one image of a number of references from which a stream of image blocks is predicted, and weighting factors from a set of weighting factors for one stream of image blocks. 2. The video decoder as defined in claim 1, whereby the reference image weighting factor unit has a second output for consideration in determining the offset corresponding to the index of a particular reference image. 3. The video decoder as defined in claim 1 includes a variable-length decoder for signal communication with the weighting-reference factor unit for providing a specific reference image to the weighting-reference factor unit. 4. The video decoder, as defined in claim 1, shall also include a motion compensation for signal communication with the reference image weighting factor for the preparation of a compensated reference image in response to the reference image weighting factor. 5. The video decoder as defined in claim number 4 includes the signal multiplier for communication with the motion compensation and the weighting factor unit - reference image for applying the weighting factor to the motion-compensated reference image. 6. The video decoder as defined in claim number 4 shall also include a signal adder for communication with the motion compensation and a reference image weighting factor for applying an offset to the motion-compensated reference image. 7. The video decoder as defined in Claim 1, where the video signal is a stream of video signal containing block transformation coefficients. 8. The video decoder as defined in Claim 1 is applicable to a two-way image predictor, in which the decoder comprises a predictive method for generating the first and second predictors from two different reference images, and an averaging method for averaging the first and second predictors together using their corresponding weighting factors to generate a single averaged predictor. 9. The video decoder as defined in claim number 8, where two different reference images, both originating from the same direction relative to the image block. 10. Method for decoding video signal data for picture blocks. The method consists of: receiving at least one reference image index containing data for the picture block, each index corresponding to a specific reference image; determining the weighting factor corresponding to each index of at least one received reference image index; retrieving the reference image corresponding to each index of at least one received reference image index; compensating for the movement of the retrieved reference image; and modifying the reference image compensated for movement by the corresponding weighting factor to create a weighted reference image. 11. The procedure set forth in claim 10 includes consideration of the determination of the corresponding offset for each index of at least one index of the acceptable reference image, and correction of the reference image that has been compensated for movement by the corresponding offset. 12. The method as defined in claim 10 involves adding a weighted motion-compensated reference image to the image block data to predict the image block. 13. The method as defined in claim 12 includes storing image blocks that are predicted to be reference images for future retrieval. 14. The method as defined in claim 10 includes the video signal data, which is a stream video signal data containing block transformation coefficients. 15. The method as defined in claim 10, where a two-way image predictor is used, involves the formation of the first and second predictors from two different reference images, the movement compensation of each of the first and second predictors, and the averaging of the first and second predictors together using their corresponding weighting factors to form a single averaged predictor. 16. The method as defined in Claim 15 includes determining the first and second offsets corresponding to each first and second predictor, and adding the first and second predictors to their first and second offsets, respectively. 17. The method as defined in claim 15 where there are two different reference images, both originating from the same direction relative to the image block.