Panoramic face and ear image stitching in multi-modal recognition

The Panorama of the face has a wider field of view, and it contains much more information of the head. After integration of human faces and ears, it can reduce computational complexity of the recognition process, and achieve a more robust non-intrusive multi-modal identification. This paper mainly focuses on building 3D face and ear panoramic image from 3D point cloud data. We extract the effective regions from the raw data, and smooth the noise of the data. Feature regions and points are located based on local maximum or minimum of shape index. Different perspectives of the face data are spliced into a 3D 180-degree panoramic image based on a modified ICP algorithm. Our method is evaluated on GavabDB face database. The experimental results and the comparison with other methods indicate the effectiveness of the proposed approach.