LPAC Research Projects

Video Scene Understanding

	Human Activity Recognition    Human activity detection and recognition		Smart Space Camera Network (SSCN)    LPAC camera network for computer vision research.
	Gait Detection    Vision-based gait is an attractive biometric because it can be observed remotely and in low resolution images/videos		Symmetry of Dance    We analyze dance sequence using five primitive symmetry types. Each symmetry detection is performed on the dance pattern extracted from the dance mocap date.

Video Tracking

	Vehicle Tracking		Real time object detection and prediction
	Efficient Mean-shift Belief Propagation for Vision tracking

Biomedical Image Analysis

	A Statistical Quantification of Human Brain Asymmetry    Constructing image index features to retrieve medically similar cases from a multimedia medical database.		Facial Asymmetry as a Biometric    We are investigating the effect of facial asymmetry measurement statistics as a biometric under expression variations.
	Non-Invasive Optical Imaging in vivo for Early Detection and Advanced Diagnosis of Cancer

Computational Symmetry

	Rotation Symmetry Group Detection    Rotation Symmetry Group Detection Via Frequency Analysis of Frieze-Expansions.		Performance Evaluation of State-of-the-Art Discrete Symmetry Detection Algorithms    Performance Evaluation of State-of-the-Art Discrete Symmetry Detection Algorithms.
	Image De-fencing    We introduce a novel image segmentation algorithm that uses translational symmetry as the primary foreground/background separation cue.		A Computational Model for Repeated Pattern Perception using Crystallographic Groups    We are developing a computational model for repeated pattern perception that is able to automatically classify a given pattern into one of the 7 frieze groups, one of the 17 wallpaper groups, or one of the 230 space groups.
	Texture Replacement in Real Images    We are developing methods to replace some specified texture patterns in an image while preserving lighting effects, shadows and occlusions.		Texture Synthesis on Near-regular Patterns    We are developing near regular texture synthesis algorithms for improved natural appearances.
	Blue Band: Multi-target analysis for marching band    This project is dealing with large variable number of interacting targets with quite low resolution. This project is a middle ground between near regular texture tracking and tracking a variable number of interacting targets.	-	-