Self-Similar Reference Masks for Video Watermark Persistence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video stream technologies face challenges in robustly embedding and detecting watermarks due to degradation during motion prediction, especially when using inter-prediction, as existing methods do not effectively compensate for reference mask drifting into slices that use inter-prediction.
Innovation Solution
A system that divides a reference mask into multiple divisions corresponding to slice divisions, determines motion vectors associated with these slices, modifies blurring kernels based on motion vectors, and convolves the reference mask divisions with modified blurring kernels to create an altered reference mask that is self-similar and less prone to degradation, enhancing watermark detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a reference mask is used for watermark embedding in video streams with inter-prediction, then watermark embedding is enabled, but the reference mask degrades and drifts into slices using inter-prediction, reducing detection reliability
Solution Approach 1:
The reference mask is divided into multiple reference mask divisions corresponding to different slice divisions in the video stream. Each division is processed independently with motion compensation, preventing the entire mask from degrading uniformly. This segmentation allows the watermark to be embedded in a stable manner across inter-predicted slices while maintaining detection reliability.
Solution Approach 2:
Motion vectors are determined and blurring kernels are modified in advance based on predicted motion before the reference mask is applied to slices using inter-prediction. This preliminary motion compensation prevents the reference mask from drifting during decoding, ensuring the watermark remains detectable and the reference mask maintains its composition stability.
2Reliability
If motion compensation is applied to prevent reference mask degradation, then detection reliability improves, but computational complexity increases
Solution Approach 1:
Motion compensation is applied locally to each reference mask division based on its corresponding slice division's motion vectors, rather than applying a single global transformation. This localized approach improves detection reliability by accounting for local motion variations while keeping computational complexity manageable through parallel processing of divisions.
Solution Approach 2:
By segmenting the reference mask into divisions that correspond to slice divisions, the computational complexity of motion compensation is distributed across multiple smaller, independent operations. Each division can be processed with simpler blurring kernel modifications, reducing the overall computational burden while maintaining high detection reliability.
3Duration of action of stationary object
If the reference mask is altered to be self-similar, then persistence during decoding improves, but the manufacturing complexity of the mask increases
Solution Approach 1:
The reference mask is transformed into a self-similar form by modifying its parameters through convolution with modified blurring kernels. This parameter transformation creates a mask that maintains its correlation over time during decoding, ensuring watermark persistence. The systematic parameter modification approach keeps the generation process manageable despite the increased complexity.
Solution Approach 2:
The self-similarity transformation is applied in advance during the reference mask generation phase, before the video stream is encoded and distributed. By performing this complex transformation preliminarily, the system ensures watermark persistence during decoding without adding real-time processing complexity during video playback or detection.
Data Source
AI summary
In one embodiment, a method including dividing a reference mask into a plurality of reference mask divisions, determining a plurality of motion vectors respectively associated with a plurality of slice divisions, wherein the plurality of reference mask divisions respectively correspond to the plurality of slice divisions, modifying a blurring kernel in accordance with the plurality of motion vectors, yielding a plurality of modified blurring kernels that are respectively associated with the plurality of slice divisions, and performing at least one action to yield an altered reference mask, including for the plurality of reference mask divisions and the plurality of modified blurring kernels: convolving a reference mask division with a weighted function of at least a modified blurring kernel associated with a slice division, of the plurality of slice divisions, to which the reference mask division corresponds.