Adaptive Interpolation Filtering for Lower-Bitrate Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video data encoding and decoding systems face inefficiencies in compressing and decompressing video data, particularly in reducing the amount of encoded data while maintaining image quality, due to the energy content in residual images and the complexity of entropy encoding processes.
Innovation Solution
The proposed solution involves predicting image regions, generating residual images, applying discrete cosine transforms, quantization, and entropy encoding, with adaptive filter selection and interpolation techniques to optimize encoding and decoding processes, allowing for efficient data compression and decompression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If traditional video encoding systems transform video data into frequency domain and apply entropy encoding, then compression is achieved, but the amount of encoded data remains large due to energy content in residual images
Solution Approach 1:
The patent applies preliminary action by performing prediction and residual generation before the main encoding process. The encoder predicts the current block from reference blocks and generates residuals, which are then transformed and encoded. This preliminary prediction step removes redundant information before compression, reducing the amount of data that needs to be encoded while preserving image quality.
Solution Approach 2:
The patent extracts the essential information by separating the prediction component from the residual component. By taking out the predictable portion (reference blocks) and only encoding the differences (residuals), the system reduces the quantity of data to be encoded. The extraction of residual energy through transformation and quantization further compresses the data while maintaining necessary image quality.
2Productivity
If entropy encoding is applied to quantized frequency domain coefficients, then compression efficiency improves, but the encoding process becomes complex
Solution Approach 1:
The patent segments the encoding process into distinct stages: prediction, transformation, quantization, and entropy encoding. By dividing the complex encoding task into manageable segments, each with a specific function, the system achieves high compression efficiency while making the overall process more manageable and implementable through standardized operations at each stage.
3Quantity of substance
If video data is compressed for efficient transmission and storage, then data quantity reduces, but decoding and reconstruction complexity increases
Solution Approach 1:
The patent applies preliminary action in encoding (prediction and residual generation) so that the decoder can perform corresponding inverse operations. The encoder's preliminary processing creates a compressed representation that, while reducing data size, maintains a structure that allows for systematic decoding through inverse transformation and addition of predicted blocks, managing decoding complexity through structured reversibility.
Data Source
AI summary
Image data encoding and/or decoding apparatus comprises an inter-image predictor configured to predict samples of a current image region, of a plurality of regions of an image in a sequence of images, with respect to a reference image region of another, reference, image of the sequence of images, according to a direction vector indicating a spatial displacement of the reference image region from the current image region; an interpolation filter configured to generate the predicted samples of the current image region by spatial interpolation of samples of the reference image region, at least in instances where the direction vector indicates a spatial displacement of a non-integer number of samples, in which the interpolation filter is configured to operate according to a set of filter parameters each defining the spatial frequency response of a set of filters providing interpolation at a spatial displacement resolution of 1/m sample, where m is an integer greater than one; and a filter response selector configured to select a set of the filter parameters from two or more candidate sets of filter parameters for use in respect of at least the current image region.