Shadow Segmentation and Augmentation Using á-overlay Models that Account for Penumbra

Michael Nielsen, Claus B. Madsen

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearch

Abstract

This paper introduces a new concept within shadow segmentation. Previously, an image is considered to consist of shadow and non-shadow regions. Thus, a binary mask is estimated using various heuristics regarding structural and retinex/color constancy theories. We wish to model natural shadows so that an augmented virtual object can cast an exact shadow. The penumbras (half-shadows) must be taken into account so that we can model the soft shadows.We hope to achieve this by modelling the shadow regions (umbra and penumbra alike) with a transparent overlay. This paper reviews the state-of-the-art shadow theories and presents two overlay models. These are analyzed analytically in relation to color theory and tangibility.
Original languageEnglish
Title of host publicationProceedings fra den 15. Danske Konference i Mønstergenkendelse og Billedanalyse
EditorsSøren I. Olsen
Number of pages10
Publisher<Forlag uden navn>
Publication date2006
Pages60-69
ISBN (Print)01078283
Publication statusPublished - 2006
EventDanske Konference i Mønstergenkendelse og Billedanalyse - København, Denmark
Duration: 24 Aug 200625 Aug 2006
Conference number: 15

Conference

ConferenceDanske Konference i Mønstergenkendelse og Billedanalyse
Number15
Country/TerritoryDenmark
CityKøbenhavn
Period24/08/200625/08/2006
SeriesDIKU Technical Report
Number06/08

Fingerprint

Dive into the research topics of 'Shadow Segmentation and Augmentation Using á-overlay Models that Account for Penumbra'. Together they form a unique fingerprint.

Cite this