Remote sensing image information extraction based on Compensated Fuzzy Neural Network and big data analytics.

Area of Science:

• Artificial Intelligence
• Remote Sensing
• Image Analysis

Background:

• Medical imaging AI and big data analytics are crucial for content-based remote sensing (CBRS).
• Current methods like Classifier-based Retrieval (CR) struggle with low-level image attributes and autonomous information retrieval.
• Existing keyword/metadata models are insufficient for complex remote sensing data analysis.

Purpose of the Study:

• To propose a novel Fuzzy Class Membership-based Image Extraction (FCMIE) technology for CBRS.
• To enhance autonomous information retrieval and overcome limitations of existing CR techniques.
• To leverage compensation fuzzy neural networks (CFNN) for improved image analysis in remote sensing.

Main Methods:

• Developed Fuzzy Class Membership-based Image Extraction (FCMIE) for content-based remote sensing.
• Utilized compensation fuzzy neural network (CFNN) to compute category labels and fuzzy membership.
• Implemented a balanced weighted distance metric, hierarchical nested structure, and cyclic similarity measure for efficient retrieval.

Main Results:

• The proposed CFNN-FCMIE model demonstrated superior performance in assessment measures like Ratio of Coverage, precision, and recall.
• Achieved a 4-5% improvement across feature vectors, sample mean, and precision-recall ratio compared to CR models.
• The method enhances remote sensing image processing and autonomous visual content retrieval.

Conclusions:

• The CFNN-FCMIE model offers a significant advancement for content-based remote sensing, improving retrieval efficiency and accuracy.
• CFNN shows broad applicability in remote sensing for feature tracking, climate forecasting, and noise reduction.
• This research provides a valuable reference for medical imaging AI and big data analytics applications.