30 October 2009 Sub-pixel mapping using Fuzzy ARTMAP neural network with fused remote sensed images
Author Affiliations +
Proceedings Volume 7498, MIPPR 2009: Remote Sensing and GIS Data Processing and Other Applications; 749834 (2009); doi: 10.1117/12.832414
Event: Sixth International Symposium on Multispectral Image Processing and Pattern Recognition, 2009, Yichang, China
Abstract
In remote sensed images, mixed pixels will always be present. Sub-pixel mapping is a technique to farther make sure the spatial distribution of all classes. The present sub-pixel mapping techniques always have a limit to the accuracy since they are based only on the soft-classified proportion data at the pixel level. In fact, supplementary information at the sub-pixel level can be used to enhance precision. In this paper, a proposed method aims to use fused imagery as supplementary information sources for the traditional sub-pixel mapping model development. The fused image with high resolution is obtained using a Gram-Schmidt spectral sharpening method, then new abundance images can replace the original data and provide more details through the Linear Mixing Model (LMM) and Fully Constrained Least Squares (FCLS) method, finally the prepared data are incorporated in a Fuzzy ARTMAP neural network. The completed algorithm is tested on a synthetic SPOT5 MS image, PAN image should be taken as a reference image. The result suggested that fine spatial resolution fused imagery can be used as a supplementary data for sub-pixel mapping, and it can get better performance than directly ANN method.
© (2009) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Ke Wu, Rui-Qing Niu, Yi Wang, Bo Du, "Sub-pixel mapping using Fuzzy ARTMAP neural network with fused remote sensed images", Proc. SPIE 7498, MIPPR 2009: Remote Sensing and GIS Data Processing and Other Applications, 749834 (30 October 2009); doi: 10.1117/12.832414; https://doi.org/10.1117/12.832414
PROCEEDINGS
8 PAGES


SHARE
KEYWORDS
Associative arrays

Fuzzy logic

Error analysis

Image fusion

Neural networks

Image classification

Image processing

Back to Top