Docket Number: 05A-930408
Inventor: Chun-Shien Lu, Chao-Yong Hsu
IP Status:TW I288356
Abstract:
Media hashing techniques have attracted much attention in content management recently due to its non-invasive property. Analogous to media hashing, there exists some synonymous terminologies including fingerprinting, digital signature, and passive/non-invasive watermarking. The common disadvantage of the existing image hashing techniques is their limited robustness against geometric distortions (For instance, resistance to rotations is restricted to very small angles).Our major contribution is to achieve robustness against extensive geometric distortions (e.g., Stirmark3.1 and Stirmark4.0). This kind of strong robustness was never achieved in the literature. Extensive results obtained from benchmark attacks have further confirmed the robustness of the proposed scheme. In addition to the robustness issue, the practical use of an image hashing system requires fast search in a large database. In this paper, we will also describe how an efficient hash database could be built to facilitate fast hash matching. In order to provide error-resilient capability in content management, we construct an error correction code (ECC)-based hash database.
Fields of Application:
Digital image management : copy detection and tracing、content authentication and multimedia database auto-indexing. We could apply this to audio and video extensively
Advantages when compared to the existing technologies:
- Wavelet-based denoising for robust mesh generation
- Image mesh-based frequency-dependent robust hash method
- Construction of error-resilient hash database
- A fast image copy detection scheme based on error correction code (ECC)
- Our invention has been applied to image copy detection and content authentication
To the best of our knowledge, the existing image hashing methods can only resist a limited number of attacks. Even the above-mentioned methods did not really deal with geometric distortions. However, our mesh-based image hashing scheme is able to efficiently resist geometric distortions. The above claim has been proven by careful statistical analysis and experimental test.
Contact Person:
對本技術有興趣,請於本處網頁廠商選項下(廠商需求與諮詢)網頁填寫資料,承辦人將跟您聯絡。
If you are interested in this technology, please fill in the information on the webpage under the manufacturer option (Apply for Technology Licensing, Material Transfer, or Sponsored Research) on the website of this office, and the undertaker will contact you
