Recovering color from black and white photographs
Восстановление цвета из черно‑белых фотографий
2010-03-01
SCID: 54.1/76qtdcqe
Discuss with AI
Bayesian analysiscolor recovery from black and white photographsexpected error estimationhyperspectral datasetsmultiple photographic sources
Figures from the paper
Abstract (AI)
This paper presents a mathematical framework for recovering color information from multiple photographic sources. Such sources could include either black and white negatives or photographic plates. This paper's main technical contribution is the use of Bayesian analysis to calculate the most likely color at any sample point, along with an expected error value. We explore the limits of our approach using hyperspectral datasets, and show that in some cases, it may be possible to recover the bulk of the color information in an image from as few as two black and white sources.
Key Findings
1
A mathematical framework is introduced to recover color information from multiple photographic sources such as black-and-white negatives or photographic plates.
2
Bayesian analysis is applied to compute the most likely color at each sample point and to provide an expected error for that estimate.
3
Evaluation on hyperspectral datasets explores the limits of the approach and demonstrates its empirical behavior.
4
In some cases, the method can recover the bulk of an image's color information from as few as two black-and-white sources.
Research Object
Multiple photographic sources consisting of black-and-white negatives or photographic plates
Research Subject
Recovering original color information (and associated expected error) from those black-and-white sources using a Bayesian mathematical framework, including limits demonstrated on hyperspectral datasets
Publication Details
Publication Date
2010-03-01
Journal
Publisher
ISSN
Access Type
Author Information
Download PDF
Subscribe to digest