【摘要】 通过卫星图像识别地面小尺度低分辨率民居屋顶截面目标,使用C4.5算法依据古民居不同年代的图像数据指纹对古民居进行分类。古民居屋顶(下称古民居)卫星图像处理形成灰度图,应用直方图与SUSAN算法精确分离出民居单元,与已知古民居位置形成不同年代"古民居图像指纹数据(组)",在冗余避免的改良C4.5分类算法中完成对民居分类与统计,实验结果 Kappa系数约为0.9107,统计结果与实际人工验证统计相符合。更多还原

【Abstract】 Through the satellite image recognition ground small scale low resolution residential roof section graphics,useing C4.5 algorithm based on ancient dwellings different eras image features of fingerprint data of ancient dwellings are classified.Researching the ancient dwellings the roof satellite image processing to form the grayscale applications histogram SUSAN algorithm accurately separate residential unit;through ancient houses are known to form the image of the different eras of ancient houses fingerprint data(group),the improvement to avoid redundantC4.5 classification algorithm to complete residential classification and statistics,experimental results Kappa coefficient of approximately 0.9107,the statistical results are consistent with the actual manual verification statistics.更多还原