A computationally efficient approach to the estimation of two- and three-dimensional hidden Markov models

Area of Science:

• Computer Vision
• Statistical Modeling
• Image Analysis

Background:

• Statistical modeling is crucial for large-scale image analysis.
• Hidden Markov Models (HMMs) are powerful tools for image segmentation.

Purpose of the Study:

• To develop a computationally efficient parameter estimation algorithm for 2-D and 3-D HMMs.
• To apply these models to satellite and volume image segmentation.

Main Methods:

• Developed a novel parameter estimation algorithm for 2-D and 3-D HMMs.
• Compared the new algorithm with existing variable state Viterbi methods for 2-D HMMs.
• Proposed a 3-D HMM for volume image modeling and segmentation.

Main Results:

• The proposed algorithm demonstrates computational efficiency for 2-D HMM parameter estimation.
• Experiments show the effectiveness of 3-D HMM for volume image segmentation using synthetic data.
• The 3-D HMM shows potential as a stochastic modeling tool for volume images.

Conclusions:

• The new parameter estimation technique is computationally efficient for 2-D HMMs.
• 3-D HMMs offer a promising approach for stochastic modeling and segmentation of volume images.