Illumination compensation and feedback of illumination feature in face detection

作者：Hong Liu, Wen Gao, Jun Miao, Debin Zhao, Gang Deng, Jintao Li · 年份：2001 · DOI：10.1109/icii.2001.983097 · 被引用次数：27 · 研究领域：Advanced Vision and Imaging、Image Processing Techniques and Applications、Video Surveillance and Tracking Methods

We present an idea of multi-method integration (MMI) to compensate variety of illumination in face detection. Some traditional methods to compensate illumination based on spatial-domain and frequency-domain are introduced. An improved algorithm LogAbout for realizing more effective illumination compensation is described. Experiments and comparisons with each scheme of multi-method integration (MMI) are presented. The experiment data show that the correctly detected rate is obviously increased after illumination compensation with the scheme MMI. This idea can solve the illumination compensation problem quickly and efficiently, which fits for detecting face automatically and tracking face in real time. The idea MMI proposed in the paper can also contribute to general image processing, object detection and recognition. In the paper, a method to feedback illumination feature is also proposed, which we think is the basis to compensate illumination partially.