Duong, H.D. and Tinh, D.T. (2013) An efficient method for vision-based fire detection using SVM classification. In: 2013 International Conference on Soft Computing and Pattern Recognition, SoCPaR 2013, 15 December 2013 through 18 December 2013.
An efficient method for vision-based fire detection using SVM classification.pdf
Download (1MB) | Preview
Abstract
In this paper, we present a new vision-based algorithm for fire detection problem. The algorithm consists of three main tasks: pixel-based processing to identify potential fire blobs, blob-based statistical feature extraction, and a support vector machine classifier. In pixel-based processing phase, five feature vectors based on RGB color space are used to classify a pixel by using a Bayes classifier to build a potential fire mask (PFM) of image. Next step, a potential fire blob mask (PFBM) is computed by using the difference between two consecutive PFM and a recover technique. In blob-based phase, for each potential blob in a potential fire blobs image (PFBI) an 7-feature vector are evaluated; this vector includes three statistical features of colour, four texture parameters and one shape roundness parameter. Finally, a SVM classifier is designed and trained for distinguish a potential fire blob are fire or fire-like object. Experimental results demonstrate the effectiveness and robustness of the proposed method. © 2013 IEEE.
Item Type: | Conference or Workshop Item (Paper) |
---|---|
Divisions: | Faculties > Faculty of Information Technology |
Identification Number: | 10.1109/SOCPAR.2013.7054125 |
Uncontrolled Keywords: | Algorithms; Color; Computer vision; Feature extraction; Fire detectors; Image processing; Pattern recognition; Pixels; Soft computing; Support vector machines; Vector spaces; Vectors; Bayes Classifier; Fire detection; Statistical feature extractions; Statistical features; Support vector machine classifiers; SVM classification; Texture parameters; Vision based algorithms; Fires |
Additional Information: | Conference code: 111424. Language of original document: English. |
URI: | http://eprints.lqdtu.edu.vn/id/eprint/10074 |