Access to SPIE eBooks is limited to subscribing institutions. Access is not available as part of an individual subscription. However, books can be purchased on SPIE.Org
Chapter 9:
Processing and Display Algorithms
Author(s): John R. Schott
Published: 2009
DOI: 10.1117/3.817304.ch9
Relatively little research on processing of polarimetric remotely sensed image data has been published. In part this is because high-resolution polarimetric image data has not been widely available for operational sensors to spur application scientists to develop and publish processing algorithms. It is also partially due to the fact that most of the algorithms developed for processing multispectral data can also be applied to polarimetric data. For example, Wolff (1990) used simple gray level thresholding of the polarization ratio to separate metals from nonmetals and Thilak et al. (2005) show that standard multiband material classification techniques can be used to classify materials using Stokes vector images of simple targets. These multiband processing methods are well treated in the literature and will not be covered here [Schowengerdt (2006), Richards (1999), and Schott (2007)]. Some image processing and display approaches particular to polarimetric image data have been developed and will be introduced here. In processing polarimetric images, the first issue to recall is that essentially all of the processing steps involve image differences that will exaggerate any misregistration between the images. Thus, careful registration of the raw filtered images before any further processing is a critical first step. Because the filtered images are highly correlated, most conventional registration methods relying on correlation are applicable and will not be addressed here [Schowengerdt (2006) and Schott (2007)]. However, the reader is cautioned that high levels of subpixel registration are required, particularly if per-pixel quantitative analysis is planned. 9.1 Display of Polarimetric Images The first possible way to view polarimetric images is to look at the raw filtered images. However, these data are so highly correlated that they tend not to accentuate the polarimetric properties of the scenes. The next step is typically to compute the Stokes vector images and to display the Stokes vector (see Fig. 9.1). The Stokes vector images can be combined using color techniques to allow simultaneous viewing of the Stokes parameters.
Online access to SPIE eBooks is limited to subscribing institutions.

Back to Top