Goesta H. Granlund

Papers

Total Citations

H-Index

About

Gösta H. Granlund is a pioneering figure in computer vision and image analysis, whose work has fundamentally shaped how machines perceive and interact with the world. His research spans attention control, multi-dimensional signal processing, and biologically inspired vision systems, with a particular focus on enabling robots to intelligently focus on relevant visual information. In his highly cited 1996 paper "Attention control for robot vision," Granlund introduced a groundbreaking method for neglecting low-level filter responses from already modeled structures, leveraging a novel filtering technique called normalized convolution. This work, which has garnered over 20 citations, demonstrated how a robot could continuously track its own moving arm while filtering out irrelevant visual data—a key step toward autonomous visual attention. Beyond this, Granlund is renowned for his contributions to hierarchical image representation and the development of the "channel representation" framework, which has influenced fields from medical imaging to autonomous systems. His interdisciplinary approach, blending signal processing with cognitive science, has left a lasting impact on both theoretical and applied computer vision, inspiring generations of researchers to build more perceptive and efficient visual systems.