No abstract

Faces is a unique region in our body that can be used as a biometric identity. Furthermore, the face between two people that have a kinship relationship may share the same face features which can be used to decide whether two people have a kinship relationship or not. In this paper, we proposed a family-aware convolutional neural network (CNN) for the visual kinship verification problem. Our proposed classifier is constructed by paralleling the state-of-the-art face recognition model and attaching two additional networks, a family-aware network, and a kinship verification network. The family-aware network weights adjusted by learning features specific to the family using deep metric learning loss while the kinship verification network use softmax loss to learn the kinship verification problem. One of the advantages of our proposed classifier is that the output of the classifier is normalized and can be represented as the probability of two images being kin or non-kin. To preserve the face recognition features extraction ability in the state-of-the-art face recognition model, we freeze the weights of the convolutional layers in the classifier for the training process. In the testing process, the family-aware network is detached to construct the final classifier. Experiments on FIW (Families In the Wild) dataset show that our proposed classifier performs better comparing with classifiers that trained without a family-aware network and the ensemble version of the classifier is comparable with several state-of-the-art methods with an average accuracy of 68.84%.

In this paper, we proposed a parallel spatial pyramid CNN classifier for image-based kinship verification problem. Two face images that compared for kinship verification treated as input for each parallel convolutional network of our classifier. Each parallel convolutional network constructed using spatial pyramid CNN classifier. At the end of the convolutional network, we use three fully connected layers to combine each spatial pyramid CNN features and decided the final kinship prediction. We tested the proposed classifier using large-scale kinship verification dataset, called FIW dataset, consists of seven kinship problems from 1,000 families. In our approach, we treated each kinship problem as a binary classification problem with two output. We train our classifier separately for each kinship problem with same training configuration. Overall, our proposed method can achieve an average accuracy of more than 60% and outperform the baseline method.

No abstract

This research designed an image encryption system that focused on securing teledermatology data in the form of skin disease images. The encryption and decryption process of this system is done on the client side using chaos-based encryption with confusion and diffusion techniques. Arnold's cat map is the chaotic map model used for confusion, while the Henon map is used for diffusion. The initial values of both chaotic maps are obtained from a 30-digit secret key that is generated using Diffie-Hellman key exchange. During Arnold's cat map generation, different p and q values are used for every iteration. On the other side, the precision of the Henon map's x and y values is 10-14. From the tests that have been done, histograms of the encrypted images are relatively flat and distributed through all the gray values. Moreover, the encrypted images have average correlation coefficients of 0.003877 (horizontal), -0.00026 (vertical) and -0.00049 (diagonal) and an average entropy of 7.950304. According to the key sensitivity test, a difference of just one number in the secret key causes big differences, as both results have a similarity index of 0.005337 (0.5%). Meanwhile, in the decryption process, that small key difference cannot be used to restore the encrypted image to its original form and generate another chaotic image with average entropies of 7.964909333 (secret key difference) and 7.994861667 (private key difference).