Hi,
I would like to know what are some possibilities to detect fire/flame in real time using opencv. I did some research on this subject, and i found people use fuzzy logic model, histograms and etc. Any information would be great.
Thanks
Another way could be to use the fact that fire actually involves motion between frames in a video input. You could start by segmenting motion first, by simply applying background subtraction and then perform color segmentation based on color histogram matching.
Just a simple idea.
The classical way to detect/classify textures would be the usage of gabor-filters together with a classifier. However, since you ask specifically about OpenCV: You could train a cascade-classifier with LBPs since LBPs are also often used for texture detection/recognition this would be worth trying. Due to the cascade of weak classifiers this should be definitely fast enough for real-time (of course the training step is not real time and takes several hours).
I just did some research on the cascade-classifier class. If i understand correctly, i have to provide at least couple thousand images of fire (correct and incorrect) to train the algorithm to detect fire/flame right?
Do you have any tutorial/examples on using cascade-classifier with LBPs?
Thanks
I can only point to the OpenCV user guide: http://docs.opencv.org/doc/user_guide/ug_traincascade.html . LBP can be selected with the 'featureType'-option.
Asked: 2013-03-03 02:09:51 -0600
Seen: 4,240 times
Last updated: Mar 04 '13
