摘要
目前的森林火灾预防主要依赖卫星监控、人工监控和区域巡场等。一种较为经济有效的方法是利用现有的用于区域巡场的监视塔,在其上设置CCD相机,定期拍摄监控照片,通过短波通讯设备将其传回指挥中心。考虑到短波通信的带宽限制,必须压缩图像,带感兴趣区域的压缩方式能在保证目标区域信息损失较小的情况下提供更高的压缩比。文章首先对图像采用分水岭算法和归一化分割,之后根据火焰的颜色特征信息提取感兴趣区域的掩模,最后在子带上以JPEG2000标准针对不同区域采用不同的压缩码率以提高整体压缩比。
Nowadays,forest fire preventions mainly compose of satellite monitoring,manual monitoring,aerial survey and so on.A more practical fireproofing surveillance tower set with CCD camera captures fire images at regular intervals and then sends them back to the command center with short-wave radio equipment.Considering the finitude of short wave bandwidth,the images must be compressed.The region of interest coding can provide higher compression rate.In this paper,the image zoning is implemented using watershed and normalized cuts.Then,the mask of the region of interest is extracted based on the features of flame.Lastly zoning is performed in the subband of images within the JPEG2000 standard to improve the calculating speed.
出处
《半导体光电》
CAS
北大核心
2015年第5期853-856,共4页
Semiconductor Optoelectronics
