¹ School of Software, Henan Institute of Engineering, ZhengZhou, 451191, China
² CHINA TOBACCO HENAN INDUSTRIAL CO., LTD., ZhengZhou, 450003, China

---

Abstract—In order to realize the recognition for dynamic facial images, the paper builds a dynamic matching model. First, this paper introduces a dynamic feature extraction algorithm of feature constraint optimization which can effectively extract the 2D dynamic facial features. Then, the paper analyzes the neurons mathematical model comprehensive expressing neurons operational mechanism and applies the model to dynamic facial feature expression. Finally, we study the learning rules how BP algorithm directs neural network, establish a corresponding mathematical model, and then use facial feature dynamic rules making dynamic features to be learned quickly and complete the recognition by the error compensation. The experiment results show that the model has a higher precision of geometric feature extraction for dynamic two-dimensional face images and better recognizes characteristic with error within 0.035mm which meets the requirements of stable, reliable, high precision and anti-interference ability etc.

Index Terms—face recognition; genetic algorithm; neural network; supervised learning

[PDF]

Cite: Hong-yan Zhang and Sheng-wei Fan, " Analysis and Simulation on Recognition Algorithm for Dynamic Facial Images," Journal of Computers vol. 8, no. 11, pp. 3004-3011, 2013.