Abstract: In infrared face recognition, eyeglasses which are the most common occluding object in facial images, have caused the loss of information of face eyes area and have seriously affected the effects of face recognition. Aiming at eyeglasses removal in face images, this paper proposes an algorithm, it detects the eyeglass regions in infrared test images and the detected eyeglass regions are replaced with an average eye template that is obtained from the average of all thermal face images without glasses. Then based on fusion of visual and thermal signatures and kernel principle component analysis, it reconstructs facial image without eyeglasses. It completes face recognition by using the classification. The experimental results show this method can easily increase face recognition accuracy and achieve good recognition effect, in the case of wearing glasses, in infrared face recognition.

Key words: thermal infrared image, glasses obstructions, kernel principle component analysis, face recognition

摘要： 在热红外人脸识别中，眼镜作为人脸图像中常见的遮挡物，造成了人脸眼睛区域信息的丢失，严重影响了人脸识别效果。针对该问题，提出了一种在热红外图像中去除眼镜的算法，对热红外图像进行眼镜检测，使用无眼镜的热红外图像的平均眼睛模板来代替有眼镜的热红外图像的眼镜区域，再基于核主成分分析算法利用可视化图像和热红外图像融合的方法，进行图像融合，获得较好的无眼镜热红外图像，通过分类识别来实现人脸识别。实验结果表明，在热红外人脸识别中，该方法在戴眼镜的情况下能够提高人脸识别的准确率和取得较好的识别效果。

关键词: 热红外图像, 眼镜遮挡物, 核主成分分析, 人脸识别