摘要
以交叉相关光谱匹配(cross correlogram spectral matching,CCSM)为基础构建土地覆盖变化强度指标,利用华北农业植被覆盖区2期不同时相的TM图像计算该地区土地覆盖变化强度图像。认为变化强度图像任意二阶邻域中像素的变化强度服从隐马尔可夫模型,用马尔可夫随机场-最大后验估计(maxium a posteriori estimation of markovrandom field,MRF-MAP)的方法从变化强度图像中提取植被变化区域。实验证明:该方法能够有效识别各种外源噪声造成的农业植被覆盖区域同物异谱的现象,可准确提取植被变化区域;但对于水体区域存在误判现象。
In this paper,a change intensity indicator of land cover based on cross correlogram spectral matching(CCSM) technique was employed to generate the change intensity image of the cropping vegetation cover area in North China between two TM images in different periods.It was first considered that the change intensity of image pixel of the two-order neighbor in the change intensity image obeyed the hidden markov random field model,and then the vegetation cover change area was extracted from the change intensity image using maximum a posteriori estimation of markov random field(MRF-MAP) model.The experiment has proved that the proposed method could precisely extract vegetation cover change and inhibit effectively the same object with different spectra due to exogenous noises in the cropping vegetation cover area.However,this method seems to perform unsatisfactorily over the water area.
出处
《国土资源遥感》
CSCD
北大核心
2012年第2期92-97,共6页
Remote Sensing for Land & Resources
基金
国家高技术研究发展计划项目(编号:2006AA120101)
国家自然科学基金项目(编号:40871191)共同资助