Optimization of Artistic Image Segmentation Algorithm Based on Feed Forward Neural Network under Complex Background...

Area of Science:

• Computer Science
• Artificial Intelligence
• Image Processing

Background:

• Traditional image segmentation algorithms struggle with noise sensitivity and complex artistic image features.
• Feed Forward Neural Networks (FFNNs) offer a promising framework for image analysis tasks.

Purpose of the Study:

• To optimize an art image segmentation algorithm using FFNNs.
• To enhance feature extraction and improve noise resilience in artistic image segmentation.

Main Methods:

• Incorporation of residual units in the encoder-decoder architecture.
• Extraction of multi-scale features using a feature pyramid module.
• Development of an improved weight adaptive algorithm for feature preservation and noise reduction.

Main Results:

• The optimized algorithm achieved accuracy and recall rates of 96.574% via 50% cross-validation.
• Demonstrated superior performance across all segmentation evaluation indexes compared to existing algorithms.
• Showcased efficient segmentation of artistic images with reduced processing time and manual intervention.

Conclusions:

• The proposed FFNN-based optimization significantly enhances art image segmentation accuracy and robustness.
• The algorithm offers a valuable reference for future research in artistic image segmentation.
• The method effectively addresses noise sensitivity and improves feature extraction for complex art images.