关键词: 平均模板, 图像纠正, 图像处理, 人脸识别

Abstract: An image preprocessing approach was presented to improve the robustness of face recognition-Image Correction By Average Template(ICBAT). ICBAT compared a face image with the average template, and constructed a weight through the difference of them,and it realized correction of the face image by adding the weight to it. Experiments show that the proposed method can greatly reduce the bright speckles and shadows of images caused by lighting and decrease the sensitivity of face recognition to the variations of facial expression and illumination. Meanwhile, this approach can increase the independence of recognition on the number of training samples.

Key words: average template, image correction, image processing, face recognition