Study on hierarchical building extraction from high resolution remote sensing imagery

You, Yu; WANG, Siyuan; Wang, Bin; Ma, Y.; Shen, Ming; Li, Xiao

doi:10.11834/jrs.20197500

Cited by 3 publications

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The third approach, object-oriented image segmentation, treats objects as independent entities using regional segmentation.This method classifies objects based on the homogeneous nature of pixels.Common algorithms include edge detection [13], thresholding [14], and region-based methods [15].Literature [16][17][18] moves from pixel to region level, using image objects to build multiscale region expressions and describe texture, reducing pixel segmentation artifacts.However, these methods need scale adjustments for optimal results and struggle to express geographic spatial relationships.When image features are small and dispersed, results may be less accurate.…”

Section: Introductionmentioning

confidence: 99%

Remote Sensing Image Segmentation for Geospatial Entities

Geng,

Lv,

Wang

et al. 2024

Preprint

View full text Add to dashboard Cite

Remote sensing image segmentation supports image interpretation.However, current methods yield results limited to segmented maps, showing only objects' boundary contours and positions.In remote sensing images, there are deeper connections between objects, such as spatial distance and topological relationships. Extracting the relationship features between objects on the basis of segmentation can significantly enhance image information. Since geographical entities contain rich attribute and spatiotemporal relationship features, which can compensate for the shortcomings of current remote sensing image segmentation, this paper proposes a remote sensing image segmentation algorithm oriented towards geographical entities. Through this method, rich and dynamic segmentation results are obtained, including three main aspects: first, segmenting the boundary contours of geographical entities using an attribute attention mechanism, extracting semantic, geometric, and visual information of entities; second, establishing a temporal attribute matrix to describe changes in the image over time; third, extracting orientation distance, topological, and interaction relationships between entities based on a semantic network model. Finally, the results obtained by this method include an additional segmentation information table based on the segmented map, which can comprehensively demonstrate the interaction relationships between entities in the image, providing new insights for remote sensing image interpretation.

show abstract

Section: Introductionmentioning

confidence: 99%