基于SMF的遥感图像纹理目标识别方法  

Standard model feature-based object recognition method for remote sensing images

作　　者：陈韶斌[1] 蔡超[1] 丁明跃[1,2] 周成平[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所,湖北武汉430074 [2]华中科技大学生命科学与技术学院,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2010年第11期33-36,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家高技术研究发展计划资助项目(2007AA12Z166)

摘　　要：提出使用标准模型特征(SMF)在遥感图像中提取和识别纹理目标的方法.在预处理阶段进行纹理区域划分,根据纹理差异将图像划分为多个可能目标和背景区域;在识别阶段,对于每个可能的目标区域,运用SMF判定区域中每一像素所属类别,以区域中大部分像素的类别作为该区域的类别,从而排除非目标区域,获得目标区域,得到识别结果.实验表明SMF是识别遥感图像纹理目标的有效特征.A method for extracting and recognizing texture-based objects in remote sensing images was described by using SMF（standard model features）.Texture regions partition was performed,and an image was divided into multiple possible object and background regions according to texture difference.In recognition phase,for each possible object region,SMF were calculated to classify the pixels of the region into object or background,and the class of the region was determined by the majority of pixels.Using this method,the non-object regions are excluded and object regions can be determined. Thus,recognition result is acquired.Experiments demonstrate that SMF is effective for recognition of texture-based objects in remote sensing images.

关 键 词：遥感图像处理 标准模型特征 视皮层前馈模型 纹理目标识别 层次化处理 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]