Pattern Recognition And Signal Processing Using Support Vect

Project Title:

Pattern recognition and signal processing using Support Vector Machines.

Project Description:

Support Vector Machines (SVMs) are a group of related supervised machine learning algorithm based on statistical learning and widely applied in pattern recognition, regression and density estimation. The rapid developments in image acquisition and storage technology has resulted in a significant growth in large and detailed image datasets, from with image processing can generate useful information which impacts on many disciplines. Compared to other methods, Support Vector Machines (SVMs) have shown a powerful capacity for both classification and feature.
One of the important applications of SVMs is fire detection in vision-based system. Unlike other fire detection methods based on particle sampling, temperature sampling, relative humidity sampling, air transparency testing, smoke analysis, and traditional ultraviolet and infrared fire detectors, vision-based system is low equipment cost, fast response, location of fire sensed directly and visibly and wildly application in various places.
This research is mainly on applying SVMs to video-based fire detection system. The aim is to increase the reliability, raise the accuracy, improve the robustness and produce good generalization.