基于超分辨率重建的低分辨率人脸识别

Low-resolution Face Recognition Based on Super-resolution Reconstruction

现有人脸识别方法主要针对高分辨率人脸,对于作战场景中获取的低分辨率人脸,已有模型的识别效果并不理想,对人脸进行超分辨率重建是处理低分辨率人脸的一个有效方法。现有研究主要关注于重建图像的视觉效果,而忽视重建后结果的识别率。针对现有需求和存在的问题,研究使用超分辨率重建的方法,提高低分辨率人脸的识别率,从而提高精准打击的成功率。提出超分辨率重建SRR模型,与现有其他算法相比,该模型在视觉效果和识别率方面具有一定的优势。

The existing face recognition methods are mainly aimed at high-resolution faces,and the recognition effect of existing models is not ideal for low-resolution faces obtained in combat scenes,and super-resolution reconstruction of faces is an effective method to deal with low-resolution faces.The existing research mainly focuses on the visual effect of the reconstructed image,but ignores the recognition rate of the reconstructed results.In view of the existing needs and existing problems,the method of super-resolution reconstruction is studied to improve the recognition rate of low-resolution faces,so as to improve the success rate of precision strike.A super-resolution reconstruction SRR model is proposed,which has certain advantages in visual effect and recognition rate compared with other existing algorithms.