期刊名称：International Journal of Image Processing (IJIP)
电子版ISSN：1985-2304
出版年度：2012
卷号：6
期号：1
页码：13-25
出版社：Computer Science Journals
摘要：The Present demands of scientific and social life forced image processing based applications to have a tremendous growth. This growth at the same time has given numbers of challenges to researcher to meet the desired objectives of either users or from solution perspectives. Among the various challenges, the most dominating areas are: reduction in required memory spaces for storage or taken transmission time from one location to other, protection of image contents to maintain the privacy and to facilitate the mechanism to identify the malicious modification if there is any either in storage or in transmission channel. Even though there are number of methods proposed by various researchers and are existed as solutions, questions are remain open in terms of quality, cost and complexity. In this paper we have proposed the concept based on neural network to achieve the quality of compression, protection and authentication all together using the ability of universal approximation by learning, one way property and one to one mapping characteristics correspondingly. With the proposed methods not only we can authenticate the image but also positions of malicious activity given in the image can be located with high precision. Proposed methods are very efficient in performance as well as carry the features of simplicity and cost effectiveness.
