Hidden Markov Models For Two-Dimensional Data
Hidden Markov models are well-known methods for image processing. They are used in many areas where 1D data are processed. In the case of 2D data, there appear some problems with application HMM. There are some solutions, but they convert input observation from 2D to 1D, or create parallel pseudo 2D HMM, which is set of 1D HMMs in fact. This paper describes authentic 2D HMM with two-dimensional input data, and its application for pattern recognition in image processing.
Unable to display preview. Download preview PDF.
- 1.Amsterdam Library of Object Images, http://www.science.uva.nl/aloi
- 2.Eickeler, S., Müller, S., Rigoll, G.: High Performance Face Recognition Using Pseudo 2-D Hidden Markov Models. In: European Control Conference (1999), http://citeseer.ist.psu.edu
- 5.Kanungo, T.: Hidden Markov Model Tutorial (1999), http://www.kanungo.com/software/hmmtut.pdf