基于线性子空间和商图像理论的人脸光照补偿

Facial illumination compensation based on linear subspaces and quotient image theory

光照是影响人脸识别率的主要因素,它已成为人脸识别技术发展的瓶颈。根据商图像理论,在所构造的低维训练集上对待识别的图像进行光照条件估计,通过加光和去光两种方法,实现了光照补偿的目的,并通过识别实验验证了其补偿效果。

Illumination is a pivotal factor in the ratio of face recognition.It is still one bottleneck preventing development of face recognition technique.Based on quotient image theory,the illumination condition is estimated on facial de-dimension illumination training set.The aim of facial illumination compensation is implemented by two basic strategies of adding light and reducing light. And the results show that this improved facial illumination compensation method is effective.