Optimizing Multiscale Continuity in Bing Maps Imagery Using Structure Transfer Techniques
This paper by Charles Han and Hugues Hoppe explores methods for achieving smooth transitions between different scales of imagery in Bing Maps. The focus is on minimizing artifacts such as ghosting and blurring through advanced techniques like structure transfer and clipped Laplacian blending. The study presents a systematic approach to downsample and compare intermediate levels of imagery, ensuring improved spatial and scale continuity between coarse and fine-scale images. The findings are significant for enhancing satellite and aerial imagery applications within mapping platforms.
Optimizing Multiscale Continuity in Bing Maps Imagery Using Structure Transfer Techniques
E N D
Presentation Transcript
Multiscale Continuity in Bing Maps Imagery (Charles Han & Hugues Hoppe) Satellite imagery Smoothtransition Google Earth:spatial discontinuities Bing Maps:scale discontinuities Aerial imagery We optimize both spatial and scale continuity: Output (1) Reduce ghosting x0 … Downsample Input xc-1 Comparison of intermediate levels: Coarse-scaleimage xc Smoothimagepyramid Structure transfer xc+1 Clipped Laplacian blending Downsample … … Fine-scaleimage xf-1 Simple linearinterpolation Our structure transfer +clipped Laplacian blending (2) Reduce blurring [Booth 125]