A Multi-Objective Decision Making Approach for Solving the Image Segmentation Fusion Problem.

Area of Science:

• Computer Vision
• Image Processing
• Computational Intelligence

Background:

• Image segmentation fusion aims to merge multiple segmentations for improved results.
• Previous methods struggled with selecting a single optimal fusion criterion.
• This limitation hindered the full exploitation of complementary segmentation information.

Purpose of the Study:

• To propose a new image segmentation fusion model using multi-objective optimization.
• To address the challenge of selecting an appropriate fusion criterion.
• To achieve improved segmentation results by leveraging complementary information.

Main Methods:

• Developed a fusion framework based on multi-objective optimization.
• Incorporated the dominance concept to combine global consistency error and F-measure.
• Utilized a hierarchical iterative relaxation strategy for optimizing the consensus energy function.
• Employed the Technique for Order Performance by Similarity to Ideal Solution (TOPSIS) for decision-making.

Main Results:

• The proposed multi-objective energy-based model effectively combines segmentation criteria.
• Demonstrated superior performance compared to classical mono-objective methods.
• Achieved better segmentation results on publicly available databases.

Conclusions:

• Multi-objective optimization offers a robust approach to image segmentation fusion.
• The developed model successfully mitigates limitations of single-criterion methods.
• This framework provides a more informative and accurate final segmentation.