Robust Fingerprinting Method for Webtoon Identification in Large-Scale Databases
Webtoon, a portmanteau of web and cartoon, denotes a cartoon that has been published on a website. Recently, webtoons have become popular in the global Internet market. Unfortunately, the copyright infringement has emerged as a new challenge resulting in illegal profit gains. Moreover, it is difficu...
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | article |
| Lenguaje: | EN |
| Publicado: |
IEEE
2018
|
| Materias: | |
| Acceso en línea: | https://doaj.org/article/4bd90b29e2ce44eeaf0dfe7c3924789d |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | Webtoon, a portmanteau of web and cartoon, denotes a cartoon that has been published on a website. Recently, webtoons have become popular in the global Internet market. Unfortunately, the copyright infringement has emerged as a new challenge resulting in illegal profit gains. Moreover, it is difficult to apply watermarking to published webtoons, because they need to be watermarked prior to publication. In order to deal with a large number of published webtoons, it is necessary to identify each webtoon using fingerprints extracted from its webtoon image. In this paper, we propose an identification framework to detect copyright infringement due to the illegal copying and sharing of webtoons. The proposed identification framework consists of the following main stages: fingerprint generation, indexing, and fingerprint matching. In the fingerprint generation stage, the translation invariant and temporally localized fingerprints are created for distortion-robust identification. An inverted indexing of the database is implemented, using the visual word clustering method and the MapReduce framework, to store the fingerprints efficiently and to minimize the searching time. In addition, we propose a two-step matching process for faster implementation. Moreover, we measured the identification accuracy and the matching time of a large-scale database in the presence of various distortions. Through rigorous simulations, we achieved an identification accuracy of 97.5% within 10 s for each webtoon. |
|---|