The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLII-3
https://doi.org/10.5194/isprs-archives-XLII-3-2353-2018
https://doi.org/10.5194/isprs-archives-XLII-3-2353-2018
30 Apr 2018
 | 30 Apr 2018

MULTI-FEATURE-MARKS BASED INFORMATION EXTRACTION OF URBAN GREEN SPACE ALONG ROAD

H. H. Zhao and H. Y. Guan

Keywords: High resolution remote sensing imagery, Urban green space along road, Spectrum features, Recognition marks, Multi-feature-marks segmentation, Information Extraction

Abstract. Green space along road of QuickBird image was studied in this paper based on multi-feature-marks in frequency domain. The magnitude spectrum of green along road was analysed, and the recognition marks of the tonal feature, contour feature and the road were built up by the distribution of frequency channels. Gabor filters in frequency domain were used to detect the features based on the recognition marks built up. The detected features were combined as the multi-feature-marks, and watershed based image segmentation were conducted to complete the extraction of green space along roads. The segmentation results were evaluated by Fmeasure with P = 0.7605,R = 0.7639,F = 0.7622.