Kim Steenstrup Pedersen

Papers

Total Citations

266

H-Index

About

Kim Steenstrup Pedersen is a leading figure in computer vision, whose research has fundamentally advanced how machines perceive and interpret visual data. His primary contributions lie in the rigorous evaluation and development of interest point detectors and image descriptors—the foundational building blocks for tasks like 3D reconstruction and object recognition. Pedersen's seminal work, "Interesting Interest Points" (2011), has garnered 194 citations, establishing a benchmark for understanding what makes a visual feature truly useful. He further addressed the critical "image correspondence problem" in "Finding the Best Feature Detector-Descriptor Combination" (2011, 44 citations), providing a systematic framework for selecting optimal pairings. His development of "Jet-Based Local Image Descriptors" (2012, 19 citations) introduced a novel, mathematically principled approach to feature representation. A key achievement is his creation of a unique dataset for evaluating interest point detectors independently of descriptors, detailed in "On Recall Rate of Interest Point Detectors" (2010, 9 citations). This work allows researchers to isolate and understand detector performance, a crucial step for improving the reliability of vision systems. Through his methodical, evaluation-driven approach, Pedersen has provided the computer vision community with both the tools and the critical understanding needed to build more robust and accurate visual algorithms.