Image Shadow Detection and Removal Based on Region Matching of Intelligent Computing

Area of Science:

• Computer Vision
• Pattern Recognition
• Image Processing

Background:

• Shadows in images degrade information for moving objects, impacting tasks like object detection and segmentation.
• Existing methods may require training or affect other image features during shadow removal.

Purpose of the Study:

• To propose an intelligent computing method for detecting and removing image shadows.
• To improve the performance of computer vision tasks affected by shadows.
• To offer a training-free shadow detection solution.

Main Methods:

• Treating each image as a small sample for analysis.
• Utilizing material matching and intelligent computing between image regions.
• Developing a method for both shadow detection and removal.

Main Results:

• The proposed method achieves direct shadow detection without requiring training.
• It ensures consistency across similar image regions during detection.
• Shadow removal minimizes the impact on other features within the shadow area.

Conclusions:

• The novel approach demonstrates promising performance in shadow detection and removal.
• It offers a significant improvement over existing advanced shadow detection methods.
• The method enhances the reliability of computer vision applications by addressing shadow interference.