Fire detection on earth's surface images in the LAB color model

Abstract

The article deals with the problem of detecting areas of images in photographs related to forest fire. In work are presented some features of data presentation for tasks related to detecting the presence of fires in images. It is shown that methods for detecting forest fires based on pixel color analysis do not always provide adequate results. Most of the known pixel color analysis techniques for fire detection use the RGB color model or combine it with the HSI model, In the work to solve the required issues, it is suggested to use the CIE LAB color model, which provides a more perceptually consistent color space compared to other color models. When using this color space, the developed decision rules and the method for detecting fires in images make it possible to determine fragments when segmenting pixels related quite accurately to fire in images. An approach to solving the detection problem using the Lab color model is presented, as it provides a better quality in terms of the accuracy of determining fires The features of data presentation for fire detection tasks on images are investigated. A method for detecting fires in images using the Lab color model is presented. Based on of computational experiments, it is shown that the proposed method provides a fairly accurate determination of fragments containing images of fires.