DERA Small Target Detection

Centre for Vision, Speech & Signal Processing

Target Detection using Motion Estimation and Data Fusion

This application represents the work by R. Crida on work package 1 of research project "Intelligent and Adaptive Scene Interpretation". It is a continuation of work started by A. Hojjatoleslami for performing blob target detection using motion estimation and data fusion.

The software described in this page forms the components of an executable RegTrack which performs tracking of targets using the developed algorithm. Details of the various parts of the algorithm are given in the class descriptions of the processing classes and instructions for using the executable are also provided below.

Notes:

Region Tracking Class Structure (docDERARegTrack)
DERARegTrack Algorithm Overview (docDERARegTrackAlgorithm)
Format of Output from DERARegTrack executable (docDERARegTrackOutput)

Subtopics:

AMMA
Applications
Image Processing
DERA Small Target Detection

Classes:

Default Classes	Description
RegBoundaryBC	Stores a region boundary as well as statistcs
RegSeedC	Stores seeds for region growing
RegOrderedGrowC	Segmentation using region growing ordered by boundary pixel intensity.
StaticTrackC	Performs static tracking stage, no temporal information
RegBoundaryC	Stores a region boundary as well as statistcs
ClusterTrackC	Performs motion cluster tracking, considers only current frame motion
FeatureTrackC	Performs history analysis of the tracked objects, ie temporal filtering
NumCostRegionBC	Implementation class for cost function used for comparing image regions.
MotionTrackC	Performs motion tracking stage
NumCostRegionC	Optimisation cost function interface for image region comparison.

Executables:

Executables	Description
DERARegTrack	Target Detection using Motion Estimation and Data Fusion

Author: Robert Crida, Created: 17/6/1998, Generated by DocEntry: March 20, 2001