Related Experiment Videos
Using color histogram normalization for recovering chromatic illumination-changed images.
1Department of Electrical Engineering, National Taiwan University, Taipei. pei@cc.ee.ntu.edu.tw
Summary
This study introduces a new image recovery method using color histogram normalization. The algorithm effectively corrects color variations caused by changing illumination using a simplified affine model.
Area of Science:
- Computer Vision
- Image Processing
- Color Science
Background:
- Image color histograms change with illumination.
- These changes can be modeled by affine transformations in R-G-B coordinates.
Purpose of the Study:
- To develop a novel image recovery method for illumination variations.
- To propose a simplified affine model for efficient color histogram normalization.
Main Methods:
- Utilizing the covariance matrix of R-G-B color histograms.
- Applying tensor theories and linear algebra for affine transformation estimation.
- Developing a simplified affine model focusing on translation, scaling, and rotation.
Main Results:
- The proposed color histogram normalization algorithm effectively recovers images under varied illumination.
- The simplified affine model outperforms the general affine model in noise-sensitive scenarios.
- The method shows effectiveness on both simulated and real images.
Conclusions:
- The novel algorithm provides a robust solution for image recovery under illumination changes.
- The simplified affine model offers improved performance and stability compared to general models.