English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Deep learning (DL) methods are widely investigated for stereo image matchingtasks due to their reported high accuracies. However, theirtransferability/generalization capabilities are limited by the instances seenin the training data. With satellite images covering large-scale areas withvariances in locations, content, land covers, and spatial patterns, we expecttheir performances to be impacted. Increasing the number and diversity oftraining data is always an option, but with the ground-truth disparity beinglimited in remote sensing due to its high cost, it is almost impossible toobtain the ground-truth for all locations. Knowing that classical stereomatching methods such as Census-based semi-global-matching (SGM) are widelyadopted to process different types of stereo data, we therefore, propose afinetuning method that takes advantage of disparity maps derived from SGM ontarget stereo data. Our proposed method adopts a simple scheme that uses theenergy map derived from the SGM algorithm to select high confidence disparitymeasurements, at the same utilizing the images to limit these selecteddisparity measurements on texture-rich regions. Our approach aims toinvestigate the possibility of improving the transferability of current DLmethods to unseen target data without having their ground truth as arequirement. To perform a comprehensive study, we select 20 study-sites aroundthe world to cover a variety of complexities and densities. We choosewell-established DL methods like geometric and context network (GCNet), pyramidstereo matching network (PSMNet), and LEAStereo for evaluation. Our resultsindicate an improvement in the transferability of the DL methods acrossdifferent regions visually and numerically.

Fine-tuning deep learning models for stereo matching using results from semi-global matching