Highlight analysis using a mixture model of probabilistic PCA

Authors:
Vladimir Bochko;Jussi Parkkinen
Affiliations:
Department of Information Technology, Lappeenranta University of Technology, Lappeenranta, Finland;Department of Information Technology, University of Joensuu, Joensuu, Finland
Venue:
ISPRA'05 Proceedings of the 4th WSEAS International Conference on Signal Processing, Robotics and Automation
Year:
2005

Citing 5
Cited 0

A physical approach to color image understanding

A physical approach to color image understanding
Mixtures of probabilistic principal component analyzers

Neural Computation
NETLAB: algorithms for pattern recognition

NETLAB: algorithms for pattern recognition
Computer Vision: A Modern Approach

Computer Vision: A Modern Approach
Parameter Estimation of a Reflection Model from a Multi-band Image

PMCVG '99 Proceedings of the 1999 IEEE Workshop on Photometric Modeling for Computer Vision and Graphics

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we propose an efficient algorithm for highlight removal in spectral images. According to the Dichromatic Reflection Model spectral data is characterized by highlight and body reflection. The considered technique is based on a mixture model of probabilistic principal component analyzers (PPCA) that is used for clustering the data into the highlight and body-reflection clusters and for computing the eigenvectors of the covariance matrices of the clusters. The K-nearest-neighbor algorithm (KNN) replaces highlight pixels with body-reflection pixels, which leads us to highlight removing in spectral images. We present the experimental results confirming the method's feasibility.