PROJECT TITLE :
Microwave Unmixing With Video Segmentation for Inferring Broadleaf and Needleleaf Brightness Temperatures and Abundances From Mixed Forest Observations
ABSTRACT:
Passive microwave sensors have higher capability of penetrating forest layers to obtain additional info from forest canopy and ground surface. For forest management, it's useful to check passive microwave signals from forests. Passive microwave sensors can detect signals from needleleaf, broadleaf, and mixed forests. The observed brightness temperature of a mixed forest will be approximated by a linear combination of the needleleaf and broadleaf brightness temperatures weighted by their respective abundances. For a mixed forest observed by an N-band microwave radiometer with horizontal and vertical polarizations, there are two N observed brightness temperatures. It is fascinating to infer four N a pair of unknowns: a pair of N broadleaf brightness temperatures, 2 N needleleaf brightness temperatures, one broadleaf abundance, and 1 needleleaf abundance. This is often a challenging underdetermined drawback. During this paper, we devise a completely unique method that mixes microwave unmixing with video segmentation for inferring broadleaf and needleleaf brightness temperatures and abundances from mixed forests. We tend to propose an improved Otsu method for video segmentation to infer broadleaf and needleleaf abundances. The brightness temperatures of needleleaf and broadleaf trees will then be solved by the nonnegative least squares solution. For our mixed forest unmixing drawback, it seems that the normal least squares resolution yields the specified positive brightness temperatures. The experimental results demonstrate that the proposed method is able to unmix broadleaf and needleleaf brightness temperatures and abundances well. The absolute variations between the reconstructed and observed brightness temperatures of the mixed forest are well inside one K.
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here
