Digital Content Marking via Windowed Glitch Patterns

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing digital audio and video fingerprinting techniques face challenges in achieving perfect invisibility or inaudibility of marks, leading to undesirable distortion and limitations in scalability, blind detection, and fine-tuning of mark insertion.

Innovation Solution

A method and device that modify data packets in digital streams by dividing segments into windows, applying a mathematical transformation to encode an internal parameter, and introducing glitches to create a unique pattern representing the receiver device's identifier, allowing for on-the-fly pattern generation and detection without reference information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If information is inserted into audio or video content to create a fingerprint mark, then the content can be identified and traced, but the content is distorted and becomes noticeable or audible

Engineering Contradiction:
ImprovetraceabilityVSAvoiddistortion
Core Design Contradiction:
ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent applies local quality by modifying only specific portions of the content rather than the entire signal. Watermark bits are embedded by selectively modifying coefficients in the transformed domain (e.g., DCT coefficients in video blocks or audio frames), allowing the mark to be hidden in localized regions where it is least perceptible while maintaining overall content quality.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent replaces direct temporal or spatial modifications (mechanical approach) with transformations in the frequency or coefficient domain. Instead of modifying raw audio samples or video pixels directly, the invention transforms the content into a different representation space where modifications are less perceptible, then applies inverse transformation to embed the watermark.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If existing fingerprinting techniques are used, then marks can be embedded in content, but the process is complex and scalability is limited

Engineering Contradiction:
Improvemark embeddingVSAvoidprocessing complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the content processing into distinct stages: transformation to a suitable domain (e.g., DCT, wavelet), coefficient selection and modification for watermark embedding, and inverse transformation. This segmentation allows each stage to be optimized independently and facilitates parallel processing, reducing overall complexity and improving scalability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The invention changes the representation parameters of the content by transforming it into a different domain where watermark embedding becomes simpler and more efficient. By adjusting transformation parameters (e.g., block size, transform type) and watermark embedding strength, the system can balance between robustness and imperceptibility while maintaining computational efficiency.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If marks are embedded in content, then identification is possible, but perfect invisibility or inaudibility cannot be achieved

Engineering Contradiction:
ImprovedetectabilityVSAvoidperceptibility
Core Design Contradiction:
ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent applies partial action by embedding watermark information in only a subset of available coefficients rather than modifying all content data. By selecting specific coefficients that are least perceptible when modified (based on their frequency content, masking properties, or importance to content quality), the system achieves detectable watermarks with minimal perceptible distortion.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The invention introduces an intermediary transformation process between the original content and the watermarked output. This intermediary step (e.g., DCT transform, wavelet decomposition) acts as a mediator that converts the content into a representation where watermark embedding is less intrusive, then transforms back to the original domain, effectively hiding the watermark from direct perception.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3066838B1Device and method to mark digital audio or audio and/or video content
Publication Date: 2018.06.13 NAGRAVISION SA
  • EP3066838B1 patent drawingFigure 1A~1C
  • EP3066838B1 patent drawingFigure 2A~4

AI summary

An aim of the present invention is to propose an improved technique of mark insertion in digital audio or audio and/or video content data by managing and minimizing the number of modification and therefore reducing the undesirable distortion of inserted marks into audio and/or video content.A receiver device including at least one processor, memories, a descrambler and decoders modules is configured to produce a sequence of elementary media content packets. The receiver device further comprises a marker unit configured to mark by modifying data in the sequence of elementary media content packets according to a pattern defined by an internal parameter associated to the receiving device. The marker unit is configured to determine, in the sequence of elementary media content packets at least one segment of a predetermined length and starting at a predetermined position in the sequence, to divide the segment into a predetermined number of contiguous windows including data of all or part of at least one elementary media content packet. Each window in the segment defined in the sequence of elementary media content packets is indexed by an index defining a position of a window in the segment. The marker unit codes the internal parameter by applying a mathematical transformation function, obtaining a resulting bits string, calculates position indices of a predetermined number of windows on the basis of all or a portion of the resulting bits string, said predetermined number of windows being lower than the number of windows in the segment, and modifies data in the windows designated by the calculated position indices. The windows including modified data and the windows including original data form within the segment of elementary media content packets a unique pattern representing the internal parameter associated to the receiving device.