| Summary: | This paper formulates a new mathematical algorithm for automated flame detection in high-speed image. The proposed algorithm is simple, unsupervised and computationally inexpensive to segment a sequence of such images into two levels which correspond to flame and background. Flame images are used as input to calculate and predict the threshold value of the image based on Otsu's method. The threshold value of the current image is corrected to compute the flame boundary. The main idea is to exploit the images' correlation in time to predict a suitable binarization threshold from the previous image. The algorithm is expected to compute the projected burnt area, which will be used to obtain the flame characteristics in a fast, robust, and reliable way. Then the proposed scheme was compared with two famous different methods and explore their effects on detection characters in scene images. It is found that the proposed method outperforms the other two prior methods in detection flame edges with respect to the flame radius versus time. © 2020 IOP Publishing Ltd. All rights reserved.
|
|---|