Ethnicity recognition (ER) is an interesting research topic in diverse fields like surveillance system, image/video understanding, investigation and so on. Presently, deep learning models become more useful in those applications. In this paper, we present a new ER model based on convolution neural network (CNN). In addition, a chaotic encryption-based blind digital image watermarking method is applied to the recognized images for security images. A cover image is employed to conceal the recognized image to protect the images from attackers or third parties. For examining the results of the presented model, an experimental analysis is carried out using VNFaces dataset which contains a set of images gathered from Facebook pages of Vietnamese people. A comparison is made with the ER-VGG model interms of accuracy. The simulation outcome indicated that the presented EG-CNN model is superior to other model on the applied images.
Deep Learning with Chaotic Encryption based Secured Ethnicity Recognition
2019-06-01
1514769 byte
Conference paper
Electronic Resource
English
Analysis of Hybrid Chaotic Image Encryption
IEEE | 2018
|Image encryption using chaotic logistic map
British Library Online Contents | 2006
|An Image Encryption Scheme Based on Cross Chaotic Map
IEEE | 2008
|