Shadow Detection in Remote Sensing Images Based on Spectral Radiance Separability Enhancement

Area of Science:

• Remote Sensing
• Image Analysis
• Computer Vision

Background:

• Shadow detection is crucial in remote sensing but challenged by features like vegetation and water bodies.
• Existing methods struggle with distinguishing shadows from dark objects and water.
• Understanding spectral properties of features is key to improving shadow detection.

Purpose of the Study:

• To develop a simple and effective shadow detection method for multispectral remote sensing images.
• To leverage physical properties of features to enhance shadow detection accuracy.
• To suppress non-shadow features while highlighting shadow regions.

Main Methods:

• Utilizing spectral properties of vegetation (brighter in NIR) and water bodies (stronger reflection in green vs. blue band).
• Developing transformation models to suppress non-shadow features and enhance shadows.
• Employing color space transformations and dominant color components to identify candidate shadows.
• Introducing normalized Color Difference Composite Index (CDCI) and Color Purity Index (CPI) for shadow separation and confidence assessment.

Main Results:

• The proposed method successfully detects shadows in multispectral images.
• The technique effectively suppresses interfering features like vegetation and water bodies.
• Experimental results demonstrate superior performance compared to state-of-the-art shadow detection approaches.

Conclusions:

• The developed method offers a robust solution for shadow detection in multispectral remote sensing.
• The fusion of spectral properties and color-based indices provides high accuracy.
• This approach advances the capabilities of automated image analysis in remote sensing.