TriMatch

New TriMatch framework fuses features for better image correspondence

Researchers have developed TriMatch, a new framework for two-view correspondence learning that improves accuracy by fusing multiple feature types. This approach combines geometric, texture semantic, and structural seman…