【摘要】：利用高空間分辨率遙感影像準確提取建筑物信息是在更精細尺度上分析、管理人類活動的重要途徑,但建筑物形態(tài)的多樣性、復雜性使得建筑物提取方法在精度、效率等方面依然面臨挑戰(zhàn)。基于高分辨率影像中建筑物所呈現(xiàn)的多種形狀及結(jié)構(gòu)特點,通過發(fā)展形態(tài)學建筑物指數(shù)并結(jié)合具有良好保邊去噪性能的引導濾波,提出了基于多結(jié)構(gòu)元素形態(tài)學建筑物指數(shù)(MMBI)的建筑物提取方法。將之應(yīng)用于具有不同特征的GeoEye-1影像建筑物提取實驗,并與MBI方法進行比較,發(fā)現(xiàn)該方法準確率達到88.2%以上,比MBI的準確率提高了約5%,且漏檢率下降近6%。結(jié)果表明:MMBI方法能夠快速、準確地提取出高分辨率影像中的建筑物區(qū)域,在建筑物的自動化提取方面具有較好的應(yīng)用前景。
[Abstract]:The accurate extraction of building information from high spatial resolution remote sensing images is an important way to analyze and manage human activities on a more precise scale. However, the diversity and complexity of building forms make the method of building extraction more accurate. Efficiency and other aspects still face challenges. Based on the features of various shapes and structures of buildings in high-resolution images, the morphological building index is developed and guided filtering with good edge preserving and denoising performance is combined. A method of building extraction based on morphological building index (MMBI) of multiple structural elements is proposed. The method is applied to the building extraction experiment of GeoEye-1 images with different characteristics and compared with the MBI method. It is found that the accuracy of this method is more than 88.2%, which is about 5% higher than that of MBI, and the rate of missing detection is reduced by nearly 6%. The results show that the MMBI method can quickly and accurately extract the building area in the high-resolution image, and has a good application prospect in the automatic extraction of buildings.
【作者單位】： 中山大學地理科學與規(guī)劃學院廣東省城市化與地理環(huán)境空間模擬重點實驗室綜合地理信息研究中心;
【基金】：國家自然科學基金面上項目“基于遙感與智能優(yōu)化方法的承災體信息提取及分布模擬研究”(41371499)
【分類號】：TP751
，

本文編號：2392110