Hash-Based Video Content Identification via Spatial-Temporal Entropy Hashing
Find Innovative SolutionsGenerate 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
Engineering 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
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.
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.
2Reliability
If exhaustive search through large video libraries is performed, then complete content identification is achieved, but search time and computational resources increase significantly
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.
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.
3Measurement precision
If spatial and temporal hash values are computed for all video sequences, then search precision is improved, but computational complexity increases
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.
Data Source
Figure 1
Figure 2
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.