HSI和区域生长结合的火灾图像分割方法

Research on Image Segmentation of Fire Based on HSI and Region Growing

森林火灾图像分割是火灾特征和识别的重要前提,其分割结果将直接影响到火灾识别的准确率。针对常用的图像分割方法进行了分析,在此基础上提出了HSI模型和区域生长结合的森林火灾图像分割方法。该方法首先将原图像转换到HSI空间,提取图像中H、S、I分量;然后在原图像中选取种子,并对其H、S、I分量图像进行区域生长;最后对各分量区域生长后的图像进行合并,最终得出分割图像。并与常用分割方法仿真结果进行了比较,试验结果表明:该算法对森林火灾分割精度高、抗扰性好且应用范围广泛,对森林火灾分割、识别具有重要意义。

The image segmentation of forest fires is the major premise of extraction and recognition of fire and its result affects accuracy of fire identification directly. Common image segmentation methods are analyzed and a new image segmentation method based on HSI and region growing is put forward. First, convert the original image into HSI space and extract H, S and I component of the image separately. Then, choose the seed in original image and do region growing to its H, S and I component. At last, combine each image of regional growth and the image segmentation image is obtained. The new method is compared with the common segmentation method. The simulation result indicates that the new method has better performance of forest fires segmentation accuracy approaching and immunity, is important for forest fn＇e segmentation and recognition.