Title :
A Spectral Color Analysis and Colorization Technique
Author :
Bochko, Vladimir ; Parkkinen, Jussi
Author_Institution :
Lappeenranta Univ. of Technol.
Abstract :
Colorization is a computerized process in which color is added to a gray-level image or movie. As the demand for colorization increases, so does the need for an automated technique. We analyze color in the spectral domain and present a spectral color-picking and colorization technique for surface color coating. A solution to the color-picking task involves principal component analysis-based learning techniques such as a mixture model of probabilistic principal component analyzers and regressive PCA. Experimental results confirm the method´s feasibility
Keywords :
image colour analysis; image resolution; learning (artificial intelligence); principal component analysis; probability; regression analysis; automated colorization technique; computerized colorization technique; gray-level image; gray-level movie; principal component analysis-based learning technique; probabilistic principal component analyzer; regressive PCA; spectral color-picking; spectral image color analysis; surface color coating; Computer graphics; Covariance matrix; Data structures; Equations; Image color analysis; Multidimensional systems; Pixel; Principal component analysis; Spectral analysis; Vectors; color; computer vision; machine learning;
Journal_Title :
Computer Graphics and Applications, IEEE
DOI :
10.1109/MCG.2006.95
