An adaptive multi-level digital watermarking method based on encoding process optimization

By employing a multi-modal intelligent analysis and coding collaboration-based multi-level watermarking method, the adaptability and robustness issues of watermarking technology in ultra-high-definition videos are resolved, achieving efficient protection in video compression, transcoding, and other processing, and ensuring the integrity and concealment of copyright information.

CN121940550BActive Publication Date: 2026-07-07HANGZHOU BAOMIHUA TECH CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN ยท China
Patent Type
Patents(China)
Current Assignee / Owner
HANGZHOU BAOMIHUA TECH CO LTD
Filing Date
2026-03-30
Publication Date
2026-07-07

AI Technical Summary

Technical Problem

Existing digital watermarking technologies struggle to achieve a balance between adaptability, coding coordination, robustness, and concealment in ultra-high-definition videos. They are unable to effectively resist video compression, transcoding, and other processing, and their single-layer protection system has weak risk resistance.

Method used

Employing multimodal intelligent analysis, the system adaptively adjusts the watermark strength, position, and type through a multi-layered watermark architecture and deep collaboration with video coding. Combined with encryption technology, it achieves differentiated optimization of the watermark during the coding process and performs intelligent processing through a reinforcement learning-driven dynamic decision-making model.

Benefits of technology

It significantly enhances the survivability of watermarks under various attack and signal processing scenarios, balances robustness and concealment, ensures that video quality is not affected, and achieves intelligent copyright protection throughout the entire process.

โœฆ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121940550B_ABST
    Figure CN121940550B_ABST
Patent Text Reader

Abstract

This invention discloses an adaptive multi-layer digital watermarking method based on encoding process optimization. The method first performs deep preprocessing on the input video stream using a pre-trained multimodal deep learning model to extract spatial complexity, temporal dynamics, and content saliency features, generating a unified feature vector. Based on the feature vector, a dynamic decision model is constructed to adaptively determine the watermark strength, embedding position, and type, achieving precise matching between watermark parameters and video content characteristics. A multi-layer watermark hierarchical embedding mechanism is employed, embedding robust invisible watermarks, fragile invisible watermarks, and dynamically visible watermarks into the video respectively, and combining chaotic encryption or hash chain encryption to enhance security. During copyright verification, an adaptive extraction algorithm recovers the watermark information of each layer, and a blockchain evidence storage system is used to complete data verification and infringement tracing. This invention constructs a complete copyright protection system encompassing embedding, encoding, extraction, and verification.
Need to check novelty before this filing date? Find Prior Art