Color-identification system using the sandglass-type neural networks

Shin Ichi Ito, Kensuke Yano, Yasue Mitsukura, Norio Akamatsu, Rajiv Khosla

研究成果: Conference article査読


A problem for the illegal copy of the digital image, which overflows on WWW, and the copyright protection of the image information becomes serious. Electronic watermark technology is being watched as that strong countermeasure. However, there is a problem that the problem that quality of output deteriorates, and the codes put inside are analyzed. In this paper, a neuro-color compression ID system with no quality of output deterioration, which sets up security for the image, is proposed. The position information of pixel chosen at random from the color image and RGB value are compressed with a proposal system by using sandglass type neural network (SNN). The information that it is compressed and the combination weight of SNN are called neuro-color compression ID. Moreover, neuro-color compression ID can be restored by using the position information of the number of pixel of SNN chosen at random. It is possible that the possible system of the prevention of an illegal copy and the copyright protection is built by making either of the position information of neuro-color compression ID, the number of pixel that it is chosen, or pixels base an image license. Furthermore, because it doesn't need to put a code inside like an electronic watermark, even the field that a demand for the information secret is severe can be used as a new watermark technology with no quality of output deterioration. Furthermore, in order to show the effectiveness of the proposed method, simulations are done. This proposed method was recognized as the patent.

ジャーナルLecture Notes in Artificial Intelligence (Subseries of Lecture Notes in Computer Science)
2773 PART 1
出版ステータスPublished - 2003 12月 1
イベント7th International Conference, KES 2003 - Oxford, United Kingdom
継続期間: 2003 9月 32003 9月 5

ASJC Scopus subject areas

  • 理論的コンピュータサイエンス
  • コンピュータ サイエンス(全般)


「Color-identification system using the sandglass-type neural networks」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。