人脸识别通常做法为:
1、提取人脸特征
2、通过计算特征相似度来判断两张人脸图像的相似程度
相似度通常用欧式距离或余弦距离
欧式距离
d i s ( X , Y ) = ∑ i = 1 n ( x i − y i ) 2 dis(X, Y) = \sqrt{\sum_{i=1}^{n}(x_{i}-y_{i})^2} dis(X,Y)=i=1∑n(xi−yi)2
def euclidean_distance(face_encodings, face_to_compare):
return np.linalg.norm(face_encodings - face_to_compare, axis=1)
余弦距离
c o s Θ = ∑ i = 1 n ( A i × B i ) ∑ i = 1 n ( A i 2 ) × ∑ i = 1 n ( B i 2 ) = A T ⋅ B ∥ A ∥ × ∥ B ∥ cos \Theta = \frac{\sum_{i=1}^{n}(A_{i}\times B_{i})}{\sqrt{\sum_{i=1}^{n}(A_{i}^2)}\times\sqrt{\sum_{i=1}^{n}(B_{i}^2)}} = \frac{A^T\cdot B}{\left \| A \right \| \times \left \| B \right \|} cosΘ=∑i=1n(Ai2)×∑i=1n(Bi2)∑i=1n(Ai×Bi)=∥A∥×∥B∥AT⋅B
def cosin_metric(x1, x2):
# single feature vs. single feature
return np.dot(x1, x2) / (np.linalg.norm(x1) * np.linalg.norm(x2))
def cosin_metric(features, x):
# all features vs. single feature
return np.dot(features, x)/(np.linalg.norm(features, axis=1)*np.linalg.norm(x))