Fire Tracking in Video Sequences Using Geometric Active Contours Controlled by Artificial Neural Network

摘要: Automatic fire and smoke detection is an important task to discover forest wildfires earlier. Tracking of in video sequences can provide helpful regional measures evaluate precisely damages caused by fires. In security surveillance applications, real-time segmentation both regions represents a crucial operation avoid disaster. this work, we propose robust tracking method for using artificial neural network (ANN) based approach combined with hybrid geometric active contour (GAC) model on Bayes error energy functional wildfire videos. Firstly, estimation function built local global information collected from three color spaces (RGB, HIS YCbCr) Fisher's Linear Discriminant analysis (FLDA) trained ANN order get preliminary pixel classification each frame. This used compute initial curves the level set evolution parameters control providing refined processed The experimental results proposed scheme proves its precision robustness when tested different varieties scenarios whether wildfire-smoke or outdoor sequences.