PAPERS
THE SECOND INTERNATIONAL CONFERENCE ON FORENSIC COMPUTER SCIENCE - ICoFCS 2007
Online ISBN: 978-85-65069-01-4, Print ISSN: 1980-1114, pp 40-47
DOI: 10.5769/C2007005 or http://dx.doi.org/10.5769/C2007005
Single Value Decomposition to Maximize the Signal-to-Noise Ratio on Digital Image
By Miguel Archanjo Bacellar Goes Telles Jr, Antonio Nuno de Castro Santa Rosa, and Paulo Quintiliano
To dowload this paper, click here.
Online ISBN: 978-85-65069-01-4, Print ISSN: 1980-1114, pp 40-47
DOI: 10.5769/C2007005 or http://dx.doi.org/10.5769/C2007005
Single Value Decomposition to Maximize the Signal-to-Noise Ratio on Digital Image
By Miguel Archanjo Bacellar Goes Telles Jr, Antonio Nuno de Castro Santa Rosa, and Paulo Quintiliano
To dowload this paper, click here.
ABSTRACT
One of the most serious problem found by users to analyze digital image is due the strong presence of salt and pepper noise which frequently masks the information of interest. Noise corrupts the signal and makes information extraction more difficult. This work brings a study of the single values matrix obtained from the Eigen Values - Eigen Vectors Decomposition of digital image. the aim of it is to maximize the signal-to-noise ratio and get a better unsupervised classification. Understanding the effects of the single values matrix over all pixels, it is proposed a denoising in order to get better results on unsupervised classification using Self Organizing Maps. To compare the performance of this filter, it is presented the classifications of the original and filtering images. In the Recognition an in the Facial Identification, however this kind of problem are common. However the technical combination of the search minimize this effect.
KEYWORDS
Eigen Values - Eigen Vectors Decomposition, unsupervised classification, Self Organizing Maps.
One of the most serious problem found by users to analyze digital image is due the strong presence of salt and pepper noise which frequently masks the information of interest. Noise corrupts the signal and makes information extraction more difficult. This work brings a study of the single values matrix obtained from the Eigen Values - Eigen Vectors Decomposition of digital image. the aim of it is to maximize the signal-to-noise ratio and get a better unsupervised classification. Understanding the effects of the single values matrix over all pixels, it is proposed a denoising in order to get better results on unsupervised classification using Self Organizing Maps. To compare the performance of this filter, it is presented the classifications of the original and filtering images. In the Recognition an in the Facial Identification, however this kind of problem are common. However the technical combination of the search minimize this effect.
KEYWORDS
Eigen Values - Eigen Vectors Decomposition, unsupervised classification, Self Organizing Maps.
To return to the "Published Papers" main page, click here.