Method of determining image classification segmentation scale threshold

Abstract

The invention discloses a method of determining an image classification segmentation scale threshold. The method comprises steps: a remote sensing image is inputted and image parameters and an initial segmentation scale are set, wherein the image parameters comprise a spectrum and a shape factor; a heterogeneity segmentation estimation algorithm is adopted to carry out N segmentations with primary circulation on the remote sensing image, wherein N is the set segmentation times; in view of a network hierarchical structure formed after segmentation of the remote sensing image, the scale layer number is counted, analyzed and calculated, and the heterogeneity local variance of each layer and the variance variations between each layer and the upper layer and the lower layer are analyzed; and a layer number with a large variance variation is acquired, the heterogeneity index of the layer is further extracted and serves as a segmentation scale threshold. In remote sensing monitoring of the geographical condition, for multi-scale high-spatial resolution satellite remote sensing data, the segmentation scale estimation can be effectively solved, a reference scale is provided, the time cost needed in the process can be effectively solved, the geographical condition remote sensing monitoring efficiency is improved, and the classification efficiency can be effectively improved.

Description

technical field [0001] The invention relates to the field of spatial image analysis, in particular to a method for determining a threshold value of an image classification and segmentation scale. Background technique [0002] my country has a vast territory and a large population, and is currently in the transition period of rapid economic development. The surface changes are meticulous and frequent, and the geographical and national conditions are huge and complex. Faced with such a situation, geographic national conditions monitoring is an important mission of spatial information science in the new era, and its implementation requires the use of air-space-ground integrated remote sensing technology and global satellite navigation and positioning technology to achieve information integration and rapid update. To a large extent, the timeliness and comprehensiveness of the monitoring objects of geographical conditions require that the earth observation has a strong ability to...

AI Technical Summary

Problems solved by technology

[0004] Image segmentation is based on the discontinuity and similarity features of image brightness values ​​as reference values, based on homogeneity or heterogeneity criteria, see figure 1 , set the optimal segmentation scale to divide the image into several sub-regions, but currently there is no accurate estimation of the optimal segmentation scale in this process, relying on empirical thresholds to set the segmentation scale to segment remote sensing data of different scales takes a lot of time Completing multiple experiments with remote sensing engineers with better skills to gain experience is the most time-consuming process in OBIA

Images