Parametric Adaptive Loop Filter Indexing for Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for processing next-generation video contents with high spatial resolution, high frame rate, and high dimensionality require significant increases in memory storage, memory access rate, and processing power, necessitating efficient coding tools to manage these demands.
Innovation Solution
A method for applying parametric adaptive loop filtering to reconstructed images, involving parsing a filter index from a bit stream to determine a filter set, deriving filter coefficients, and applying these coefficients to a reconstructed image obtained from a predicted signal and a difference signal, which reduces the amount of bits required for adaptive loop filtering and improves image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional adaptive loop filtering is used for next-generation video contents, then image quality can be maintained, but the amount of bits required and processing complexity increase significantly
Solution Approach 1:
The patent changes the parameter representation from full filter coefficients to a compact form using a base filter set index and residual coefficients. This parameter transformation reduces the bit requirements while maintaining the ability to represent the same filter characteristics, thereby resolving the contradiction between maintaining image quality and reducing bit requirements.
Solution Approach 2:
The patent extracts the essential filter characteristics into a predefined filter set that can be selected by index, separating the major filter behavior from the fine-tuning residuals. This extraction allows the bulk of filter information to be transmitted efficiently through indexing rather than full coefficient transmission, reducing bit requirements while preserving image quality.
2Measurement precision
If conventional adaptive loop filtering is applied to reconstructed images, then filtering accuracy can be achieved, but device complexity and processing power requirements increase
Solution Approach 1:
The patent segments the filtering process into two distinct stages: selecting a base filter set from a predefined collection (low complexity) and applying residual coefficient adjustments (fine-tuning). This segmentation reduces processing complexity by breaking down the complex filter coefficient generation into simpler, more manageable operations while maintaining filtering accuracy.
Solution Approach 2:
The patent performs preliminary action by pre-defining a set of common filter characteristics that can be selected by index. This preliminary preparation eliminates the need to compute full filter coefficients from scratch during decoding, reducing processing complexity while maintaining the ability to achieve accurate filtering through the combination of base filters and residuals.
Data Source
AI summary
Disclosed in the present invention are a method for encoding/decoding an image and a device for same. Particularly, a method for decoding an image may comprise the steps of: parsing a filter index from a bit stream outputted from an encoder; determining a filter set using the filter index, from among a plurality of filter sets belonging to a predetermined filter bank; deriving a filter coefficient based on a coefficient for filtering, defined by the filter set, which has been determined by the filter index; and applying filtering to a reconstructed image obtained using a prediction signal and a differential signal, by using the derived filter coefficient.