Classification of Clouds in Satellite Imagery Using Adaptive Fuzzy Sparse Representation

Area of Science:

• Meteorology
• Remote Sensing
• Computer Vision

Background:

• Satellite cloud imagery is crucial for weather forecasting and climate monitoring.
• Cloud pattern analysis is a recent research focus.
• Satellite data presents challenges like fuzziness, uncertainty, noise, and outliers, impacting traditional classification methods.

Purpose of the Study:

• To propose an adaptive fuzzy sparse representation-based classification (AFSRC) method for satellite cloud classification.
• To address the limitations of traditional methods in handling noisy and uncertain satellite cloud imagery.
• To improve the accuracy, stability, and adaptability of cloud classification.

Main Methods:

• Introduced an improved fuzzy membership function with adaptive parameters (attenuation rate, critical membership) to handle image fuzziness and uncertainty.
• Combined the improved fuzzy membership with sparse representation-based classification (SRC) to optimize training dictionary atoms.
• Developed an adaptive fuzzy sparse representation classifier for cloud classification.

Main Results:

• The proposed AFSRC method demonstrated improved accuracy in satellite cloud classification.
• The method exhibited strong stability and adaptability when tested on FY-2G satellite cloud images.
• The classification process showed high computational efficiency.

Conclusions:

• The AFSRC method effectively overcomes the challenges of fuzziness and noise in satellite cloud imagery.
• This approach offers a more reliable and accurate solution for meteorological applications.
• The study highlights the potential of adaptive fuzzy sparse representation for advanced cloud analysis.