A new flame segmentation algorithm based color space model

Flame image segmentation is an important prerequisite for fire characteristics and recognition, The segmentation results will affect the accuracy of fire identification. This paper proposes a flame segmentation algorithm merging 2-dimension Otst and seed region growth on HSI color space. After many texts, it confirmed that the new flame segmentation algorithm doesn't only overcome interference the light source reflection area and smoke area in fire image, but also quickly segment the relatively complete suspected fire area.