Research On Method Of Multi-scale Landscape Pattern Extraction Based On HJ-1A HSI Image

Remote sensing is a core technology for acquiring large-scale and multi-scale landscape pattern data in research of landscape ecology, and multi-scale landscape pattern data is the base for a series of landscape ecology scale study including scale effect, scale domain, rescaling and so on. Therefore, developing the study of methods for multi-scale landscape pattern extraction based on remote sensing is of important theoretical and applied value. For a long period of time, in the study of landscape pattern with remote sensing, it is only the spatial resolution that is concerned about, ignoring the spectral resolution. To solve the problem, the basis for choosing remote sensing image is extended from a single spatial resolution to multi-basis including spatial resolution, spectral resolution and information entropy of remote sensing image, according to which HJ-1A HSI data is introduced into landscape pattern study, and the extraction methods of landscape pattern based on HJ-1A HSI data is also be studied. Besides, grain size is often treated as the pixel size, which leads to treating grain transformation as the pixel-based aggregation and decomposition. In this paper, the limitation of the grain transformation method based on pixel aggregation is analyzed, and grain transformation method based on no-changing pixel, according to which the GIS neighborhood analysis is brought into study of landscape pattern grain transformation. Based on the solution of data source and analyzing method, the programs of multi-scale transformation of landscape pattern based on HJ-1A HSI data and GIS neighborhood analysis is proposed. The experimental results show that the programs of multi-scale transformation of landscape pattern proposed in this paper can effectively improve the accuracy and reasonableness of the multi-scale landscape pattern through the expansion and improvement from choosing remote sensing data and grain transformation method, which in order to propose a new way for thinking for study of multi-scale landscape pattern.