Face Recognition System Based on Improved LVQ Neural Network Algorithm
Hao Sun · 2022
With the development of computer technology, face recognition has become one of the hot research topics today. Aiming at the current problems of face orientation recognition such as low recognition accuracy and influence by lighting conditions, this paper starts from the principle of LVQ algorithm, by establishing the improved LVQ algorithm model and applying it to face feature matching, realizing the session to establish 400 face images as training set and 40 as test set, and deriving the recognition accuracy of the improved LVQ algorithm model as 93%. The experiments compare the BP algorithm and SVM algorithm in four aspects: training set accuracy, training set loss function, test set accuracy, and recognition effect on different data sets, and verify that the LVQ algorithm has a greater advantage in accuracy and stability.