Super resolution land cover mapping technique and its applications

Teacher

Yong Ge

Title

Ph.D. Professor

Orgnization

Institute of Geographic Sciences and Natural Resources Research , Chinese Academy of Sciences (IGSNRR, CAS)

Length

0:59:40

Abstract

Super resolution mapping (SRM) technique decomposes each mixed pixel into a fixed number of subpixels based on a zoom factor and it then assigns these subpixels to specific LULC classes. In contrast to common classifiers, SRM can produce a LULC map with finer spatial resolution than the original moderate- or low-spatial-resolution input image. Thus, SRM offers a solution to the tradeoff between the spatial resolution of a sensor and its spectrum. It can process mixed pixels in low-, moderate-, and high-spatial-resolution imagery and thus save on the cost of obtaining images with higher spatial resolution and SRM can be applied to low- or moderate-resolution images acquired in an earlier phase to produce LULC maps with spatial resolution consistent with maps produced from high-spatial-resolution images acquired during the current phase. It is therefore attractive to use SRM to derive finer-resolution LULC maps from coarser-resolution remote sensing images. We developed a series of SRM techniques and verified the effectiveness of the techniques through the experiment. Further, we discussed the applications of SRM results, for example to investigate the consistency of the classification results from SRM and higher spatial resolution remote sensing image, the zoom factor setting for SRM. We developed the corresponding evaluation indexes and discussed the applicability of the SRM techniques in the real applications.