摘要: There are two main approaches for face recognition with variations in lighting conditions. One is to represent images features that insensitive illumination the first place. The other approach construct a linear subspace every class under different Both of these techniques successfully applied some extent recognition, but it hard extend them variant facial expressions. It observed highly sensitive expression variations, which result changes both conditions and expressions difficult task. We propose new method called Affine Principal Components Analysis an attempt solve problems. This extract representation warps this space achieve better separation. proposed technique evaluated using databases variable more than 90% accuracy by only one sample image per class.