Open AccessImproving image clarity using local feature dimension
Author(s) -
Lowe Thomas
Publication year - 2015
Publication title -
iet image processing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.401
H-Index - 45
eISSN - 1751-9667
pISSN - 1751-9659
DOI - 10.1049/iet-ipr.2014.0642
Subject(s) - raster graphics , feature (linguistics) , measure (data warehouse) , artificial intelligence , computer science , fractal dimension , computer vision , dimension (graph theory) , fractal , point (geometry) , pixel , feature vector , scale (ratio) , mathematics , pattern recognition (psychology) , geometry , data mining , mathematical analysis , geography , philosophy , linguistics , cartography , pure mathematics
This study presents an alternative method of displaying vector and raster graphics which provides greater visual clarity than standard methods. Rather than rasterising lines and points by shading them with a pixel thickness, shade is interpreted as an intensity per length and per point, respectively; generically per fractal measure of the geometric feature. Integrating these shades through supersampling provides a generic shading method that is independent of screen resolution, supersample size and feature dimension. By using a fractal measure that is local in both space and scale, the author's method generalises to arbitrary features and so is extendable to raster images where no feature is truly sub‐two‐dimensional. The resulting images exhibit details that are lost to standard rasterisers. Their system can be seen as enabling a sliding scale between a photographic view and a diagrammatic view of the same data.