1 February 2013 in Information Sciences
1 January 2012 in Journal of Electronic Imaging
Chaos maps and chaotic systems have been proved to be useful and effective for cryptography. In our study, the two-dimensional logistic map with complicated basin structures and attractors are first used for image encryption. The proposed method adopts the classic framework of the permutation-substitution network in cryptography and thus ensures both confusion and diffusion properties for a secure cipher. The proposed method is able to encrypt an intelligible image into a random-like one from the statistical point of view and the human visual system point of view. Extensive simulation results using test images from the USC-SIPI image database demonstrate the effectiveness and robustness of the proposed method. Security analysis results of using both the conventional and the most recent tests show that the encryption quality of the proposed method reaches or excels the current state-of-the-art methods. Similar encryption ideas can be applied to digital data in other formats (e.g., digital audio and video). We also publish the cipher MATLAB open-source-code under the web page https://sites.google.com/site/tuftsyuewu/source-code.
1 April 2014 in Information Sciences
25 December 2017 in IEEE Transactions on Pattern Analysis and Machine Intelligence
This paper concerns the problem of facial landmark detection. We provide a unique new analysis of the features produced at intermediate layers of a convolutional neural network (CNN) trained to regress facial landmark coordinates. This analysis shows that while being processed by the CNN, face images can be partitioned in an unsupervised manner into subsets containing faces in similar poses (i.e., 3D views) and facial properties (e.g., presence or absence of eye-wear). Based on this finding, we describe a novel CNN architecture, specialized to regress the facial landmark coordinates of faces in specific poses and appearances. To address the shortage of training data, particularly in extreme profile poses, we additionally present data augmentation techniques designed to provide sufficient training examples for each of these specialized sub-networks. The proposed Tweaked CNN (TCNN) architecture is shown to outperform existing landmark detection methods in an extensive battery of tests on the AFW, ALFW, and 300W benchmarks. Finally, to promote reproducibility of our results, we make code and trained models publicly available through our project webpage.