根据小波域中各图像之间的相关性,提出基于小波变换的多层级图像配准算法。在小波变换域利用位图切割完成多层级匹配,有效地减少了人脸图像中旋转、光照等噪声影响,保持了较高的配准精度。实验结果表明,重建的人脸图像具有较高的识别性能。
Firstly,the face images were segmented into several bitmaps by grayscale value,secondly,wavelet transformation was done on the bitmaps,and registration was finished.Noise such as rotation,light,noise,etc.,were weakened.Registration was done efficiently.Experimental results show that the method presented performs better than the traditional algorithm.