Related Experiment Video
Updated: Apr 14, 2026
09:17
Using Retinal Imaging to Study Dementia
Published on: November 6, 2017
22.5K
Illumination normalization of face image based on illuminant direction estimation and improved Retinex
Jizheng Yi1, Xia Mao1, Lijiang Chen1
1School of Electronic and Information Engineering, Beihang University, Beijing, 100191, China.
Plos One
|April 24, 2015
Summary
This study introduces a novel illumination normalization method for face images, enhancing face and facial expression recognition. The technique effectively normalizes illumination without requiring training or 3D models, improving image quality.
Area of Science:
- Computer Vision
- Image Processing
- Biometrics
Background:
- Illumination normalization is a critical challenge in face recognition and facial expression analysis.
- Existing methods often require extensive training or complex 3D models.
Purpose of the Study:
- To develop an effective illumination normalization technique for face images.
- To improve the performance of face and facial expression recognition systems.
- To propose a method that does not rely on training or 3D face models.
Main Methods:
- The proposed method divides face images into local regions to calculate edge information.
- It selects optimal regions to determine illuminant direction using error and constraint functions.
- The Retinex algorithm is enhanced by optimizing the surround function and histogram stretching for improved dynamic range.
Main Results:
- The method successfully achieves illumination normalization on face images.
- Experimental results on the extended Yale face database B and CMU-PIE demonstrate superior normalization compared to existing techniques.
- The approach avoids the need for training data or 3D face/reflective surface models.
Conclusions:
- The developed illumination normalization technique is effective and robust.
- This method offers a significant improvement over current approaches for face image preprocessing.
- The technique has broad applicability in face recognition and facial expression recognition tasks.
