Feature Point Detection in Image Morphing Based Steganography
Qiangfu Zhao, Yutaro Minakawa, Yong Liu, Neil Yuwen Yen · 2013
In today's society, information hiding is becoming a key mechanism to protecting one's secrets. In our study, we have proposed several new information hiding methods based on image morphing. To synthesize a large number of virtual but natural images through morphing to cover different kinds of secrets, it is necessary to propose some efficient and effective method for automatic feature point detection. For this purpose, we proposed a method based on iterative evolutionary algorithm (IEA) earlier. This method, however, is not efficient because it is too time consuming. In this paper, we propose a neural network (NN) based method for feature point detection. The basic idea is to detect a neighborhood of a specified feature point, and use the center of the neighborhood as the estimated position of the feature point. The performance of the proposed method is verified through experiments.