Invisible Watermarking for Screen-Capture Communication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing screen-capture communication methods face challenges such as obtrusive communication, limited message size, and device-to-device synchronization issues, particularly in ensuring complete transmission and retrieval of larger messages and correcting perspective distortions during camera alignment.
Innovation Solution
The method involves embedding a hidden message in the frequency domain of video frames using digital watermarking, correcting perspective distortions through template matching, and using control information for message segmentation and sequencing to enable transmission of large messages across multiple frames without visible artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If invisible watermarking is used to hide messages in video frames, then message visibility is improved (messages become invisible), but message size capacity deteriorates (limited message size per frame)
Solution Approach 1:
The patent divides a large message into multiple smaller message segments, each of which can be embedded in separate video frames. This segmentation allows the system to transmit messages larger than the capacity of a single frame by distributing the message across multiple frames, thereby resolving the contradiction between message invisibility and message size capacity.
2Quantity of substance
If multiple frames are used to transmit larger messages, then message size capacity is improved, but device synchronization challenges worsen (devices may go out of sync)
Solution Approach 1:
The patent embeds control information in each video frame that contains instructions for assembling the message segments in the correct sequence. This preliminary action of embedding assembly instructions ensures that even if devices experience slight synchronization variations, the receiver can correctly reconstruct the complete message by following the embedded control information, thereby maintaining reliability while enabling larger message transmission.
3Manufacturing precision
If camera is held steady for capture, then perspective distortion is reduced, but ease of operation deteriorates (difficult to align devices)
Solution Approach 1:
The patent employs template matching algorithms that automatically detect and correct perspective distortions in the captured video frames without requiring manual device alignment. The system performs self-service by autonomously identifying the message regions and applying geometric transformations to correct distortion, thereby eliminating the need for users to precisely align devices while maintaining high perspective correction accuracy.
4Object-affected harmful factors
If frequency domain watermarking is used, then message hiding quality is improved (no visible artifacts), but device complexity worsens (complex processing required)
Solution Approach 1:
The patent replaces traditional spatial domain watermarking methods with frequency domain watermarking techniques. This substitution transforms the watermarking process from direct pixel manipulation to frequency-based embedding, which produces no visible artifacts in the transmitted video frames. Although this increases processing complexity, it achieves superior message hiding quality by operating in the frequency domain where human visual perception is less sensitive.
Data Source
AI summary
The present disclosure provides for methods, apparatus and computer readable media related to screen-capture communication based on invisible watermark. An aspect of the disclosure provides for a method including inserting one or more templates into one or more frames of a video for correcting frame perspective of the one or more frames captured by a device. The method further includes writing a message into the one or more frames of the video and displaying the video. According to a second aspect, a second method is provided. The second method includes capturing one or more frames of a video displayed on a device comprising a hidden message. The method further includes locating one or more templates and correcting frame perspective of the captured frames based on the templates. The method further includes extracting the message from the one or more frames of the video.