A method and system for integrated geometric and semantic processing of multi-source remote sensing satellite images

Abstract

The invention discloses a multi-source remote sensing satellite image geometry and semantics integrated processing method and system. The method is different from the traditional scheme of performing geometric precision correction on remote sensing images first, and then performing semantic segmentation and information extraction, and proposes a method that includes semantic information Extraction, automatic geometric fine correction assisted by semantic information, and semantic information optimization are three steps. Firstly, from the standard scene images, the cloud, water surface, ice and snow, cloud shadows, artificial buildings and other terrain information that have a great influence on the geometric precision correction are preliminarily extracted, and then with the assistance of these information, interference is eliminated to realize fully automatic remote sensing images. Geometric precision correction, uniform light and color, seamless mosaic, image synthesis and other processing, and finally in the high-precision multi-source composite image, extract richer semantic information and target information, and obtain ultra-large-scale multi-source composite image and its corresponding semantics Maps and thematic maps of land types.

Description

technical field [0001] The invention belongs to the field of surveying and mapping science and technology, and relates to a multi-source remote sensing satellite image geometric and semantic integrated processing method and system. Background technique [0002] The workflow of the early multi-source satellite image geometric fine correction and semantic segmentation processing system generally first obtains high-precision multi-source remote sensing satellite synthesis through processes such as image matching, block adjustment, orthorectification, image uniformity, and mosaic. Image, and then obtain semantic information such as target information, ground object coverage information, etc. from the synthetic image through intelligent means or artificial means. But in this kind of processing system, the role of semantic information in the synthesis of multi-source remote sensing satellite images has not been played. [0003] In the absence of semantic information, the high-pre...

AI Technical Summary

Problems solved by technology

However, in this type of processing system, the role of semantic information in the synthesis of multi-source remote sensing satellite images has not been played.

[0003] In the absence of semantic information, the high-precision synthesis of multi-source images will encounter the following problems: ① image matching points may be distributed on unstable features such as clouds, water surfaces, ice and snow, which reduces the accuracy of geometric fine correction; ② automatic matching to obtain The control points may be distributed on "non-ground surfaces" such as buildings, and such control points cannot automatically obtain accurate elevations from the DEM; ③ Clouds, cloud shadows, ice and snow, etc. 4. In the absence of building and tree coverage information, the mosaic line may pass through such obstacles, resulting in dislocation of the images on both sides of the mosaic line

These problems, on the one hand, reduce the quality of automatic multi-source remote sensing synthetic image production (such as reducing geometric accuracy, causing halo phenomenon in fused images, causing image color cast after uniform light and color, etc.), and on the other hand, make some work only It can only be completed through a lot of manual intervention (such as bypassing obstacles such as buildings and trees by manually editing mosaic lines)

Images