对于人脸图像识别中光照变化的影响,传统的解决方法是对待识别图像进行光照补偿,先使它成为标准光照条件下的图像,然后和模板图像匹配来进行识别。为了提高在光照条件大范围变化时,人脸图像的识别率,提出了一种新的可变光照条件下的人脸图像识别方法。该方法首先利用在9个基本光照方向下分别获得的9幅图像来构成人脸光照特征空间,再通过这个光照特征空间,将图像库中的人脸图像变换成与待识别图像具有相同光照条件的图像,并将其作为模板图像;然后利用特征脸方法进行识别。实验结果表明,这种方法不仅能够有效地解决人脸识别中由于光照变化影响所造成的识别率下降的问题,而且对于光照条件大范围变化的情况,也可以得到比较高的正确识别率。

To eliminate the effects of the changes of lightings,the conventional approaches firstly produce a compensation based face image under standard illumination from the input image,and then match it with the face templates in a database.In this paper,a novel face recognition approach for different illumination conditions is proposed.The proposed method generates a new gallery under the same lighting conditions with an input image instead of transforming the input image.First,a face illumination eigenspace is constructed from nine face images of the same person taken from nine basic directions respectively.And then,the face images in the face image database are transformed into the rendered images taken under the same illumination with the input image based on the face illumination eigenspace.Finally,the input image is matched with the newly rendered images which are regarded as a new gallery,to perform the recognition task.The experimental results show that the proposed approach can efficiently eliminate the effects of different illuminations,and has a high recognition rate in the illumination conditions with remarkable changes.