基于云计算的嵌入式人脸识别系统建构与研究

Cloud Computing-based Embedded Face Recognition System Construction and Research

摘要:

针对嵌入式系统运算、存储资源有限等缺陷,采用先进的FaceCore人脸检测开放平台构建一套基于云计算的嵌入式人脸图像处理系统；通过在高性能嵌入式平台接入图像采集设备采集图像信息,然后将复杂的图像处理、检测、识别等算法转移到云平台进行,实现人脸特征值检测、人脸特征点检测、人脸相似度检测、人脸匹配等功能,减少了嵌入式系统的本地运算负担,降低了嵌入式人脸检测、识别系统的软硬件成本；最后通过对多个不同年龄、性别的人进行人脸检测识别、人脸和嘴的位置检测以及图像中人脸个数检测,分别得到检测时间的平均值；实验结果验证了构建基于云计算的嵌入式人脸识别实时系统的可行性,系统的稳定性、可靠性,并且与以前直接在本地进行计算做比较,具有较高的实时性,为未来嵌入式系统进行实时的图像处理提供了一种更有效的解决方案。

Abstract:

As for the defects of embedded systems computing and limited storage resources,I built a embedded face processing system based on a set of cloud computing by using the advanced FaceCore open platform.Firstly,capture the image information by image capturing device connected to the embedded platform. Then by transforming the complicated image processing,detection and recognition algorithms to the cloud platform,realize facial feature value detection,facial feature detection,face similarity detection,and face matching functions. Thirdly,reduce the burden on the local operation of embedded systems and reduce the cost of hardware and software for embedded face detection,identification systems.At last,we get an average detection time from detecting the position of human face,mouth and the number of faces in the image with different age,sex,recognition of people,the experimental results verify system stability,reliability and the feasibility of the construction of the embedded real-time face recognition system based on cloud computing,it has the high real-time performance compare to that on the local before ,it provides a more effective way for the embedded system of real time image processing in the future.