Adaptive neuro-heuristic hybrid model for fruit peel defects detection
1Institute of Mathematics, Silesian University of Technology, Kaszubska 23, 44-100 Gliwice, Poland.
Summary
This study introduces an adaptive method fusing a novel neural network architecture with heuristic search for image analysis. The approach precisely detects areas of interest, such as peel damages, improving decision support systems.
Area of Science:
- Computer Science
- Artificial Intelligence
- Image Processing
Background:
- Machine learning fusion enhances decision support systems by integrating diverse features.
- Adaptive methods are crucial for optimizing information processing in complex systems.
Purpose of the Study:
- To develop an adaptive method combining a novel neural architecture and heuristic search.
- To precisely detect areas of interest in images, specifically peel damages.
Main Methods:
- Image segmentation for simplified processing.
- Development of an Adaptive Artificial Neural Network (AANN) architecture.
- Fusion of AANN with heuristic search for area-of-interest detection.
Main Results:
- The proposed method successfully identifies pixels corresponding to peel damages.
- Experimental results demonstrate the efficacy of the fused approach.
- Adaptive learning improved classification accuracy.
Conclusions:
- The fusion of adaptive neural networks and heuristic search offers a powerful approach for image analysis.
- This method enhances the precision and efficiency of decision support systems.
- The adaptive nature of the AANN is key to its performance.