Authors : C. Emmy Prema and S.S. Vinsley
Abstract: In this study the performance of fire pixel segmentation technique is analysed in RGB, YUV, CIE L*a*b* and YCbCr color spaces using statistical parameters of the image. Various rules are formed in each color space by defining color and texture feature of the fire and also research is carried out to segment the fire centre region. The true fire pixels including fire flame region and fire centre region are effectively segmented from the background in YCbCr color space than the other color spaces. The fire pixel segmentation method using static features of fire (Color and texture) is less complex than dynamic features of fire like motion, flickering, rate of growing etc. and can be incorporated in the situation when the camera is moving.
C. Emmy Prema and S.S. Vinsley, 2016. Performance Analysis of Statistical Parameter Based Fire Pixel Segmentation in Various Color Spaces. Asian Journal of Information Technology, 15: 2023-2036.