Hash-Based Video Content Identification via Spatial-Temporal Entropy Hashing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for identifying video sequences in broadcasting and audiovisual content systems are inefficient, particularly when metadata is absent or unreliable, and require exhaustive searches through large libraries of video content.

Innovation Solution

A method and apparatus that utilize spatial and temporal hash values derived from entropy measures of pixel value differences within and between frames to efficiently locate matching video content by ordering and indexing data, allowing for rapid search and identification of specific sequences within large libraries.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If metadata comparison is used to identify video sequences, then identification accuracy is improved, but the method fails when metadata is absent or unreliable

Engineering Contradiction:
Improveidentification accuracyVSAvoidmethod availability
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent creates a copy of the video sequence metadata in the form of fingerprint data derived from actual video frames. Instead of relying on original metadata that may be absent or unreliable, the system generates fingerprint copies from video content itself, ensuring identification can proceed even when original metadata is unavailable.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent introduces fingerprint data as an intermediary between the video sequence and identification process. Rather than directly comparing original metadata, the system uses fingerprint data derived from video frames as a mediator that bridges the gap when metadata is unreliable, enabling identification through content-based comparison.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If exhaustive search through large video libraries is performed, then complete content identification is achieved, but search time and computational resources increase significantly

Engineering Contradiction:
Improvecontent identification completenessVSAvoidsearch time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent segments the video library search process by organizing video sequences into groups based on their fingerprint data characteristics. Instead of performing exhaustive linear search through all sequences, the system divides the search space into manageable segments using fingerprint-based indexing, allowing faster location of candidate sequences that match the query.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary organization of video library data by pre-computing and storing fingerprint data for all video sequences in the library. This preliminary action creates an indexed structure that enables rapid retrieval of candidate sequences during search operations, eliminating the need for exhaustive real-time analysis of every sequence.

Inventive Principle:
Principle #10Preliminary action

3Measurement precision

If spatial and temporal hash values are computed for all video sequences, then search precision is improved, but computational complexity increases

Engineering Contradiction:
Improvesearch precisionVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts only the essential spatial and temporal characteristics from video frames to create fingerprint data, rather than processing all video sequence information. By taking out only the relevant features (spatial relationships and temporal changes) and converting them to hash values, the system achieves high search precision with reduced computational complexity compared to full sequence analysis.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP3161722B1Hash-based media search
Publication Date: 2024.03.06 GRASS VALLEY LTD
  • EP3161722B1 patent drawingFigure 1
  • EP3161722B1 patent drawingFigure 2
  • EP3161722B1 patent drawing

AI summary

A method of identifying an item of video content involves providing a spatial hash value and a temporal hash value for each image in a video collection. Each hash value is based on a measure of the entropy in differences between pixel values.A table of the pair of hash values against timecode is created and ordered according to one of the hash values. A search for a given pair of hash values can then be confined to that part of the table that matches the first value.