Research On The Key Technology Of Face Recognition For Welcome Robot

With the aging of the population and the continuous pursuit of people's quality of life,it is a trend to use service robots with service robots for completing various types of work.As one of the most important service-oriented robots,welcome robot needs to treat different guests differently.This requires that it not only needs to store a large number of personnel information but also respond quickly and accurately.Therefore,the key issue is the design of real-time detection and recognition technology with high recognition accuracy.To solve the above problems,this paper first introduced the current situation of facial recognition technology,and designed the face detection and recognition algorithm with C++ language.Then,these algorithms were evaluated and optimized by different face databases and practical scene.Finally,this paper developed a real face recognition system,and the experimental results demonstrated the effectiveness of the proposed system.The research work in this paper can be summarized as follows:(1)Considering the real time and accuracy of face detection algorithm,a face detection method based on theAda Boostalgorithm is proposed.In order to improve the Ada Boostmethod for false detection,we utilize the ellipse model to reduce false positive.The test results in CMU database and the actual scene show that the proposed face detection method can effectively reduce the false detection rate of facial image.(2)In practical applications,the face posture changes greatly,and the detected face image usually contains some background area.To solve this problem,the supervised descent method is used to locate 49 key points of a face image.Then,the face deflection angle is calculated according to the coordinates of the key points.According to the affine transformation function,the correction of face image is completed.The experimental results verify the effectiveness of the face corrected method.(3)When the LBP operator directly calculates the whole face image,the LBP feature is not effective to characterize the difference between different people.This paper proposed a face recognition method by fusing the LBP features around 49 key points of face images,and it is more discriminative than the single LBP features.The experimental results show that the proposed method can effectively improve the recognition accuracy.