Research Interests By WenHung Liao 9272007 Focus of
Research Interests By Wen-Hung Liao 廖文宏 9/27/2007
Focus of VIPL Research l Multimedia signal processing, including: l l l Video processing Image analysis Audio signal processing A common thread: human perception of the physical world. (cognitive multimedia processing) Put more emphasis on content analysis than compression/multimedia format/standards.
Video Processing l l l l Semantic analysis of sports video: swimming style classification Real-time video effects: compressed-domain Non. Photorealistic rendering (NPR) Face detection and tracking: (see also DDPlay. Cam) Facial expression analysis using local appearance information Video-based human computer interaction: Magic Mirror Video analysis in sleep studies Motion tracking and analysis using real-time stereo Skin color segmentation using achromatic features
Swimming Style Classification l Objective: to classify swimming motion into four styles, namely, l l Backstroke (仰式) Breaststroke (蛙式) Butterfly (蝶式) Free style (自由式) using video recorded with an above-water camera.
Demo
Compressed-Domain NPR Original video With oil-paint effect
Face Detection and Tracking
Facial Expression Analysis
Image Analysis l Biometrics: l l l l Face recognition under low illumination Recognition of human faces with glasses Automatic caricature generation (2 D and 3 D) Textured-image-based CAPTCHA (Completely Automated Public Turing Test to Tell Computers and Humans Apart: Examples) Image-based CAPTCHA Attention-based personal photo manager Eye tracking Recognition of Hand-drawn Geometric Objects
2 D Caricature: Some Results + =
3 D Caricature
3 D Caricature: Illustration
Textured-image-based CAPTCHA l Static Patterns: l Dynamic
Image-based CAPTCHA l Easy l Hard
Gaze Trajectories
Gaze Path Comparison
Audio Signal Processing l l l Voice User Interface based on Voice. XML Personalized Information Broadcasting System Birdcall recognition News audio evaluation and analysis Snoring analysis
Comments/Questions?
- Slides: 20