Extraction of architectural information from urban green space baced on high resolution remote sensing image

This paper presents a novel architecture-extraction method in urban green space based on high resolution remote sensing imagery. The method introduces the knowledge of architecture features into extraction process and makes full use of spectral and textural characteristics, thus successfully discriminates spectrally similar architectures from green space and solves the problem of inaccurate statistics of urban vegetation. First of all, spectral response of each object is analyzed to conduct Normalized Difference Vegetation Index (NDVI) extraction. Then Gray Level Co-occurrence Matrices (GLCM) is used to extract architectural information according to textural features. Moreover, we extract more accurate architectural information in vector format in terms of their geometric characteristics. This method has successfully extracted most architectural information in the green space of the study area; the result has demonstrated its high accuracy and significance in practice.