Edge Detection Comparison of Hybrid Feature Extraction for Combustible Fire Segmentation: A Canny vs Sobel Performance Analysis

Abstract

Fire is a dangerous environmental problem; putting the fire out in an incident area cannot be done without fire prevention equipment. Fire prevention equipment means any device, appliance, or machinery intended for use, and examples of this are fire extinguisher, fire alarm, and smoke alarm. After a fire has been dealt with, it takes a huge amount of time for an investigation to conclude the cause of the incident. Therefore, this study proposed an algorithm to detect fire and extract fire features using different image processing. This paper showed different types of algorithms in image processing and tried multiple edge detection techniques to determine which are best to apply in fire detection using image processing. It uses the RGB (Red, Green, and Blue) Color Model and HSV conversion to extract features of fire. Edge detection applying Sobel which has two separate directions the vertical and the horizontal. Canny edge detection that spots edges with noise suppressed at the same time. Canny also detects a wide variety of edges in images. With the help of these algorithms, in case of disaster especially fire, the growth of it will be so much easier to detect using this image processing and algorithm. Based on the testing, it proves that using Canny edge detection getting 98% accuracy more accurately than the Sobel edge detection technique. The algorithm of feature extraction of fire was validated and has a validating f-score of 0.98 using Canny edge detection and segmentation.