Shape, orientation and size recognition of normal and ambiguous faces by a rotation and size spreading associative neural network
Shape, orientation and size recognition of normal and ambiguous faces by a rotation and size spreading associative neural network
复制标题
通过旋转和尺寸扩展关联神经网络识别正常和模糊脸部的形状、方向和尺寸
DOI：
10.1109/ijcnn.2002.1007524
复制
发表时间：
2002
期刊：
影响因子：
--
通讯作者：
K. Takase
中科院分区：
文献类型：
--
作者：
K. Nakamura;K. Arimura;K. Takase
在本研究中，我们开发了一种旋转和大小扩展的关联神经网络（RS-SAN网络）来实现人脸的视图不变识别。在此基础上，开发了一种扩展的RS-SAN网络，实现了基于视图的模糊人脸识别。RS-SAN网络正确地识别了不受方向和大小影响的脸型、不受形状和大小影响的脸型、不受形状和方向影响的脸型。扩展的RS-SAN网络将每个模糊的脸识别为两个不同的脸，这取决于它是直立的还是倒置的，而不管它的大小。
In the present study, we developed a rotation and size spreading associative neural network (RS-SAN net) to accomplish view invariant recognition of human faces. Then we developed an extended RS-SAN net to accomplish view dependent recognition of ambiguous faces. The RS-SAN net correctly recognized face shape irrespective of its orientation and size, the face orientation irrespective of its shape and size, and the face size irrespective of its shape and orientation. The extended RS-SAN net recognizes each ambiguous face as two different faces depending on whether it is upright or inverted, irrespective of its size.