Pattern-Based Image Compression for Random Sub-Block Access

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image data compression methods require decompressing entire blocks to access individual values, which is inefficient for graphics processing units that randomly access small blocks of data, and caching decompressed blocks does not significantly improve efficiency.

Innovation Solution

A pattern-based compression method that divides image data into sub-blocks, encodes each sub-block based on its pattern and unique values, and utilizes pointers for matching sub-blocks to reduce redundancy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If existing image data compression methods are used, then data is compressed to reduce memory transfer, but the entire block must be decompressed to access individual values which is inefficient for random access

Engineering Contradiction:
Improveamount of dataVSAvoidaccess efficiency
Core Design Contradiction:
Quantity of substanceVSEase of operation

Solution Approach 1:

The patent divides the image data into smaller sub-blocks (e.g., 2x2 pixels) rather than treating the entire block as a single unit. Each sub-block can be independently accessed and decompressed, allowing random access to individual pixels or small regions without decompressing the entire block. This segmentation enables efficient random access while maintaining compression benefits.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a hierarchical structure with super-blocks containing multiple sub-blocks, adding a dimensional layer to the compression organization. This allows the system to navigate through compressed data using a tree-like structure where individual sub-blocks can be accessed independently, transforming the traditional flat block decomposition into a multi-level hierarchical system that supports random access.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Loss of time

If caching decompressed blocks is implemented, then access time may be reduced, but memory bandwidth consumption increases and power usage rises

Engineering Contradiction:
Improveaccess timeVSAvoidpower consumption
Core Design Contradiction:
Loss of timeVSUse of energy by moving object

Solution Approach 1:

Instead of caching entire decompressed blocks (excessive action), the patent implements a system where only the necessary sub-blocks are decompressed and cached based on actual access patterns. This partial decomposition and caching approach reduces memory bandwidth consumption and power usage while still providing fast access to frequently accessed regions through the L1 cache.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The patent implements a dynamic compression/decompression system where the behavior adapts to access patterns. The system can switch between compressed and decompressed states based on whether random access or sequential access is detected, allowing the memory subsystem to optimize its behavior dynamically rather than using a fixed caching strategy.

Inventive Principle:
Principle #15Dynamics

3Manufacturing precision

If higher quality rendering algorithms are used on fast GPUs, then rendering quality improves, but memory bandwidth becomes a limiting resource

Engineering Contradiction:
Improverendering qualityVSAvoidmemory bandwidth utilization
Core Design Contradiction:
Manufacturing precisionVSProductivity

Solution Approach 1:

The patent performs compression of image data in advance, before it needs to be accessed by the rendering pipeline. By pre-compressing the data and organizing it in a hierarchical structure with sub-blocks, the system prepares the data in an efficient format that reduces memory bandwidth requirements during rendering operations, allowing higher quality algorithms to run without bandwidth constraints.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3889910B1Pattern-based image data compression
Publication Date: 2025.12.03 IMAGINATION TECH LTD
  • EP3889910B1 patent drawingFigure 1
  • EP3889910B1 patent drawingFigure 2
  • EP3889910B1 patent drawingFigure 3

AI summary

Methods and compression units for compressing a two-dimensional block of image element values. The method includes: dividing (204) the two-dimensional block of image element values into a plurality of sub-blocks of image element values; identifying (208) which pattern of a plurality of patterns is formed by the image element values of a first sub-block of the plurality of sub-blocks; and forming (210) a compressed block of image element values by encoding the first sub-block in the compressed block of image element values with: (i) information identifying the pattern, and (ii) the image element values of the first sub-block forming the pattern; determining from a mask a set of one or more sub-blocks of the plurality of sub-blocks that have one of one or more predetermined relationships with the first sub-block; determining whether each sub-block in the set of one or more sub-blocks matches the first sub-block; and in response to determining that each sub-block of the set of one or more sub-blocks matches the first sub-block, encoding each sub-block of the set of one or more sub-blocks in the compressed block of image element values with information identifying that the sub-block belongs to a set of sub-blocks that match the first sub-block.