Fuzzy Cluster Algorithm of Multi-dimensional Dynamic Primitives for Automatic Face Recognition Based on Multi-center Matching
Junyi Tang · 2019
Automatic face recognition is an important technology giving a machine the capacity to recognize a person's identity through the computer vision. At present, there is a main problem of automatic face recognition, shown as following: the semantic features extracted by primitives are redundant and the recognition accuracy of faces with sparse characteristics is low. Therefore, this thesis puts forward the fuzzy cluster algorithm of multi-dimensional primitives for automatic face recognition based on multi-center matching. It designs the algorithm aiming at the fuzzy initial center problems of cluster during picture segmentation. By constructing a sparse construction of multi-center parallel clustering, and simultaneously converging various semantic features for many feature semantics, the algorithm improves the time-effect of recognition. Additionally, multi-center cluster reduces the feature quantity of face recognition. Through the simulation verification, the algorithm of the thesis can effectively improve the recognition speed and the accuracy.