This paper presents a new Line Enhance Active Contour model (LEAC model) to detect the flame front boundaries from high speed Planar Laser Induced Fluorescence (PLIF) images. The model first enhances the crack region and the gradient of flame front by the Line Enhance Filtering algorithm, and then extracts flame front boundaries and crake edges from the enhanced PLIF images using the CV model. We compared the LEAC model with the classical CV model on different PLIF images. Experimental results show that our model can detect the flame front accurately, and it has good performance on detecting the exact edges of long and narrow crack regions, where the classical CV model can not detect completely. Moreover, it is insensitive to image noise and curve initialization.