Video Watermarking Error Concentration Metric
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video watermarking techniques fail to detect highly noticeable artifacts such as ragged contours and flicker caused by slight edge shifts in modified blocks, which are not captured by current objective metrics.
Innovation Solution
A method that evaluates the distribution of pixel errors within a block to assess error concentration, discarding blocks where errors are concentrated on a few pixels, thereby avoiding ragged contours and flickering artifacts, using an error concentration metric that compares the normalized entropy or percentile-based error concentration against a target value.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If standard distortion metrics (RMSE, MAD) are used to evaluate watermarking modifications, then the global distortion level is captured, but localized concentrated errors producing ragged contours and flicker are not detected
Solution Approach 1:
The patent segments the error evaluation process by dividing the block into individual pixel error assessments. Instead of computing a single global metric, the method calculates error values for each pixel and analyzes their distribution pattern, identifying whether errors are concentrated or dispersed. This segmentation enables detection of localized artifacts while maintaining computational feasibility through efficient error mapping and statistical analysis.
Solution Approach 2:
The patent transitions from evaluating error magnitude in one dimension (global distortion values like RMSE) to analyzing error distribution across spatial dimensions. By examining the spatial arrangement and concentration of errors across pixels, the method detects ragged contours and flicker artifacts that scalar metrics miss, adding a spatial distribution dimension to the evaluation process.
2Reliability
If motion vectors are modified to embed watermarks, then robustness against compression is improved, but perceptual fidelity deteriorates due to blocking artifacts and ragged contours
Solution Approach 1:
The patent implements a feedback mechanism where error distribution analysis results directly influence the selection of modifiable blocks. Blocks exhibiting concentrated error patterns indicative of ragged contours or flicker are fed back into the selection process and excluded from watermarking, while blocks with acceptable dispersed error patterns are selected. This feedback loop ensures robust watermark embedding while maintaining perceptual quality.
Solution Approach 2:
The patent changes the evaluation parameter from global distortion magnitude to error distribution characteristics. By analyzing how errors are distributed across pixels rather than just their overall magnitude, the method identifies blocks where modification would create unacceptable visual artifacts and adjusts the watermarking strategy accordingly, selecting blocks that maintain both robustness and perceptual fidelity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for determining a modifiable block among a plurality of candidate blocks in a watermarking scheme is disclosed. The method comprises obtaining an error block with respect to a block value and a modified block value for a candidate block among the plurality of candidate blocks; determining an error concentration metric for the error block; and determining that the candidate block is a modifiable block by comparing the error concentration metric to a concentration value. A device and a watermarking method implementing the disclosed method are further described.