改进的阈值分割法在林火图像分析上的应用

Application of Improved Threshold Segmentation Method in Forest Fire Image

采用巴特沃思高通滤波器进行频域滤波,使用傅里叶函数变换方法消除所采集林火图像中的烟雾等噪声,在数学形态学对林火图像预处理的基础上,对火焰的动态特征进行了提取和识别。通过计算出林火图像的火焰面积、火焰圆形度和火焰质心偏移距离,并且经过试验数据证明,该算法可以有效的区分干扰源和火焰,精确的判断出是否有火险发生。

The Butterworth high pass filter was used in frequency domain filtering. The Fourier Transform method was used to eliminate the noises about smoke in the forest fire images. Based on the mathematical morphology in the forest fire images preprocessing,the dynamic characteristics of the flame was extracted and identified. The area of flame,the circularity of flame and the centroid displacement of flame were calculated. The experiment results show that the proposed algorithm can efficiently distinguish flames from the background noises,and the decision whether the fire hazard occurs can be made accurately.