Human Ear Recognition Based on Convolutional Neural Network
Yin C. Hu · Journal of North University of China · 2015
The human ear biometrics characteristics and recognition method was analyzed,and a new method was proposed based on convolution neural network(CNN).The classification and recognition was made by using twice extractions of ear feature.In the conditions of different angles and different dimensions,the bigger embedding dimension was,the more retained feature information was after dimensionality reduction,and the greater the recognition probability was;the dimension change had no influence on CNN recognition.Experimental results show that this method can maintain good recognition probability at changing posture of human ear,and has good robustness at changing the recognition angle of ears.