Abundance of vegetation plays an important role in urban ecosystem, urban planning and development. Traditional classification methods on remote sensing data by assigning each pixel membership in one, and only one have the primary shortcomings of their inability to accommodate spectrally mixed pixels in gradational land covers. The traditional classification methods are giving way to spectral mixture analysis (SMA) gradually which is better in acquiring quantitative information for specific land covers. Vegetation fraction, in a general way, is defined as the areal fractions of vegetation within each pixel. This paper, besides introducing the traditional technique of SMA, discusses the improvement of traditional technique from the aspects of data noise removal, least-squares solution with constraining sum of endmembers fractions to unit, pixel purity index and the selection of endmembers. LSMA is tested further with the Shanghai city as an example. Unmixing pixels with root mean square (RMS) error less than 0.02 accounts for the proportion of 98.5%. The spatial distribution of vegetation is corresponding to actual situation. Then we conclude that: the improved LSMA is appropriate for estimating quantitative vegetation fraction and the technique will be widely applied in urban environment. I. INTRODUCTION
[1]
P. Switzer,et al.
A transformation for ordering multispectral data in terms of image quality with implications for noise removal
,
1988
.
[2]
C. Small.
Multitemporal analysis of urban reflectance
,
2002
.
[3]
C. Small.
Estimation of urban vegetation abundance by spectral mixture analysis
,
2001
.
[4]
C. Pieters,et al.
Remote geochemical analysis : elemental and mineralogical composition
,
1993
.
[5]
Conghe Song,et al.
Spectral mixture analysis for subpixel vegetation fractions in the urban environment: How to incorporate endmember variability?
,
2005
.
[6]
Yosio Edemir Shimabukuro,et al.
The least-squares mixing models to generate fraction images derived from remote sensing multispectral data
,
1991,
IEEE Trans. Geosci. Remote. Sens..
[7]
R. Kauth,et al.
The tasselled cap - A graphic description of the spectral-temporal development of agricultural crops as seen by Landsat
,
1976
.
[8]
J. Boardman,et al.
Mapping target signatures via partial unmixing of AVIRIS data: in Summaries
,
1995
.