Satellite image time series data semantic segmentation method based on state space model

Abstract

This invention discloses a semantic segmentation method for satellite image time series data based on a state-space model, belonging to the field of intelligent remote sensing image processing technology. The method includes the following steps: S1, acquiring satellite remote sensing image time series and preprocessing them to obtain a preprocessed sequence; S2, extracting multi-temporal features using a spatial feature coding network; S3, obtaining scale features for spatial modeling; S4, obtaining the spatial modeling result; S5, obtaining the final temporal modeling features; S6, determining the pixel-level semantic segmentation result based on the spatial modeling result and the final temporal modeling features. This invention effectively improves the spatial consistency and temporal discriminative ability in satellite image time series semantic segmentation by introducing a semantically guided spatial feature modulation and temporally adaptive feature modeling method. While ensuring computational efficiency, it improves the accuracy and stability of the segmentation results, exhibiting good robustness and practical value.

Images