Biomedical and Information Processing Subgroup

This subgroup applies signal processing and machine learning techniques to aerospace, biomedical, multimedia, data storage, social-networking, and telecommunication fields.

Biomedical and Information Processing Subgroup

This subgroup applies signal processing and machine learning techniques to aerospace, biomedical, multimedia, data storage, social networking, and telecommunication fields. We have five main research topics:

  1. The first is fault detection for turbine jet engines, where normal signatures of steady-state gas path parameters are used as a reference to find and classify abnormal behavior during transient (non steady-state) engine operation.
  2. The second topic is automated mitochondrial segmentation, where machine learning techniques are used to extract texture features of morphological characters for accurate sequence partitioning.
  3. In environmental sound recognition, we focus on developing a robust ensemble learning model for the integration of features and recognition the sound content.
  4. In the study of recommendation systems for social media networks, we propose a hierarchical Bayesian model to integrate social network structures with item content-information for improved item suggestions.
  5. The last topic is energy management in large-scale parallel storage systems, where we propose an algorithm to not only minimize storage energy consumption, but also control the impact of energy saving techniques on service delays.

Biometrics Subgroup

This subgroup mainly focuses on research related to visual quality, such as image (video) quality assessment in HD (Ultra-HD) and 3D contents. The research topics is closely related to the visual quality indices (VQI) applied in HEVC coding, 2D-3D conversion, and depth-based image rendering.

Computer Vision and Scene Analysis Subgroup

The research of this subgroup is related to recognition and classification over biometrics, graphics and compound images. There are five key topics: age group classification, facial recognition, latent fingerprint identification, 3-D mesh recognition, and text detection in compound images.

Visual Quality and Perceptual Coding Subgroup

This subgroup focuses on semantic reasoning of images taken in various circumstances. Our research is to judge the contents of images and perform image retrieval, multi-view recognition, and image segmentation.