A two-branch cloud detection algorithm based on the fusion of a feature enhancement module and Gaussian mixture model

Area of Science:

• Earth and Space Sciences
• Computer Science
• Remote Sensing Technology

Background:

• Accurate cloud detection is crucial for optimizing remote sensing (RS) data utilization.
• Existing algorithms struggle with identifying challenging cloud types like edge and broken clouds.

Purpose of the Study:

• To develop an improved cloud detection method for remote sensing images.
• To enhance the identification of edge and broken clouds.

Main Methods:

• Utilized Himawari-8 satellite channel data.
• Combined a feature enhancement module with a Gaussian mixture model (GMM).
• Employed Laplacian operator for spectral feature enhancement of cloud edges and broken clouds.

Main Results:

• The proposed method demonstrated promising consistency with visual interpretation.
• Achieved superior accuracy metrics compared to Random Forest (RF), K-Nearest Neighbors (KNN), and standard GMM.
• Successfully improved the detection of edge and broken clouds.

Conclusions:

• The novel feature enhancement and GMM combination method offers high-precision cloud detection capabilities.
• This approach has significant potential for advancing remote sensing applications.
• The method effectively addresses limitations of existing cloud detection algorithms.