My Research

I am interested in advanced biometric algorithms in terms of iris recognition, face recognition, heterogeneous image synthesis and recognition. I am committed to tacking biometric recognition problems in complex unconstrained situations, such as large pose variations and extreme lightings. Some works have achieved satisfactory performances in real scenarios, and been successfully applied in mobile phones, public security, etc.

I am interested in most tasks of low-level vision, like super-resolution, deblurring, denoising, inpainting, deraining, anti-makeup, face aging and cross-spectral hallucination. Some of the works have achieved promising performances in diverse datasets and benchmarks. We have already focused on image enhancement in real-world scenarios like surveillance, and alleviated the challenges of down-stream computer-vision applications.

I am interested in developing unsupervised learning methods and generative models, including information theory learning, domain adaptation, Variational Autoencoder and Generative Adversarial Networks. The proposed algorithms achieved advanced distribution quality and visual appearance. Specifically, I am trying to conquer the essential issues in high-resolution synthesis. Some of the works have been successfully applied to novel view synthesis, portrait manipulation, and image-to-image translation.