Scalable Video Compression for Hybrid Machine Human Vision

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current video compression technologies fail to efficiently compress videos for both machine vision and human vision, as they do not adequately account for the distinct characteristics required for each domain, leading to suboptimal performance in machine-to-machine communication applications.

Innovation Solution

A scalable video compression structure is proposed, which uses an adaptive loop filter with coefficients derived through feature domain minimum error and task error minimum error methods, classifying coding tree units into significant and insignificant groups based on pixel importance and edge direction, and performing filtering accordingly to enhance encoding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional video compression is used, then encoding speed is maintained, but compression efficiency for both machine vision and human vision deteriorates

Engineering Contradiction:

Improveencoding efficiencyVSAvoidcompression performance

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The image is divided into coding tree units that are further segmented into sub-blocks, allowing different filtering operations to be applied to different segments based on their specific characteristics and importance for machine vision tasks

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Different filter coefficients are derived and applied to different coding tree units based on their importance for machine vision tasks, enabling optimized compression performance for each local region rather than using a uniform approach

Inventive Principle:

Principle #3Local quality

2Reliability

If adaptive loop filtering is applied to all coding tree units, then compression performance improves, but processing time increases

Engineering Contradiction:

Improvecompression performanceVSAvoidprocessing time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

Instead of applying complex adaptive loop filtering to all coding tree units, the method applies filtering selectively to only those coding tree units that are important for machine vision tasks, reducing processing time while maintaining compression performance for critical regions

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

Coding tree units are classified into important and less important groups before filtering is applied, allowing the system to prioritize processing time for regions that matter most for machine vision while using faster processing for other regions

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20240244199A1Encoding/decoding method for purpose of scalable structure-based hybrid task

Publication Date: 2024.07.18 KWANGWOON UNIVERSITY INDUSTRY ACADEMIC COLLABORATION FOUNDATION

AI summary

The present invention proposes a scalable-based video compression structure in a video compression technology for supporting a hybrid task. In an adaptive loop filter step of an encoder of a layer for a machine task, a coding tree unit may be classified into a coding tree unit significant group and a coding tree unit insignificant group, and, for the coding tree unit significant group, filter coefficients may be derived by a feature domain minimum error method and a task error minimum error method.