An integrated pattern
recognition approach –
Directional Edge Analytical Model (DEAM) is presented. Remote
sensing and image processing with pattern
recognition are significant for collecting information.
Many different techniques, invariant to different types of geometric transformations, contextual conditions and task demands, have been developed for recognizing patterns and objects. Most of current pattern recognition work has emphasized non-remote objects and has lacked special handling with the properties of remote sensing images. In addition, Relational structures are particularly suitable for representing complex patterns and for providing matching solutions; this area of research is still at very early stage.
Using
directional edge as description of relational features, DEAM provides supervised models for remote sensing images with matching procedures that determine patterns. Novel methods using adaptive threshold detectors and relational feature descriptors are designed according to the special properties of remote sensing images. DEAM analyzes the connection of functional edges in an object, and then it builds the analytical model by using directional edges as feature descriptors. It utilizes “many to many relations” with “relation of relations” to reduce redundant features and to increase the accuracy of a training set. After feature relational mapping, the corresponding descriptors of the model and the target will assist researchers in identifying the objects.
DEAM demonstrate superiority in quantifying structural relationships and in extracting features that are desired for many pattern recognition applications.
COPYRIGHT BY
JUNFENG GU,
School of Computing,
The University of Southern Mississippi
2007-09-06
Email Authors: junfeng.gu@gmail.com