Robotics Research Technical Report: Shape Description With a Space Variant Sensor: Algorithms for Scan-Path, Fusion and Convergence, Over Multiple Scans (Classic Reprint) - Hardcover
Synopsis
Unlock how space-variant sensing reshapes scene blending and scan strategies for sharper, lighter vision systems.
The book explores how a retina-inspired sensor with nonuniform resolution can fuse information from multiple views to form a coherent, high-detail image. It introduces an Attentional scanpath approach that selects fixation points to maximize captured detail and describes blending methods that preserve high-resolution regions while simplifying less-detailed areas. The work also defines convergence metrics to measure how quickly a fused representation approaches the original scene, with practical guidance for robotic and machine vision applications.
Grounded in contour-based imagery, the text shows how curvature and boundary-angle representations drive focus decisions, and it discusses data compression benefits and real-time considerations for space-variant imaging systems.
- How space-variant resolution affects image formation and data handling
- A simple attentional algorithm to choose successive fixation points
- Blending techniques to fuse multiple views into a single accurate contour
- Metrics for convergence and practical notes for real-time robotic use
Ideal for readers interested in machine and robotic vision, especially those curious about how humanlike vision concepts can inform sensor design and image processing.
"synopsis" may belong to another edition of this title.
