Face Image Smart Capture System Using Embedded Device

Abstract

At present, as an important means of identity authentication, face image recognition has been widely used in many fields. However, during obtaining a face image, it is often necessary for the person being captured to adjust his positions of face or a staff help to manually adjust the cameras position repeatedly. With this, the face images captured may meet the authentication requirements. For solving these problems, an intelligent face image capture system based on embedded device is designed. The hardware mainly consists of a control module with STM32F103C8T6 as CPU, and a two-degree-of-freedom pan tilt zoom camera (PTZ) with two steering gears and a digital camera. The two-scale fitting inverse combination active appearance model (AAM) algorithm is used for face image detection and location. Then the steering gear is controlled to adjust the camera’s direction according to the relationship between the center of the face image and the center of the camera’s field of view (FOV) to complete the face image capture automatically. Experiments show that the designed system meets the authentication requirements and has the good real-time performance and stability.