|
The rapid advancement of technologies in electronics and information, computer and communication, artificial intelligence is changing our life. Various intelligent devices are emerging, such as smart TV, smart phone, intelligent monitoring, intelligent toys etc. With face recognition that more and more intelligent devices analysis function. This is because a face image contains large amount of human information such as the subject’s age, gender, identity, emotion, etc. Compared with other modes of human-computer interaction, it has the attractive property of user friendly and convenience.On the other hand, with the development of technology of natural language processing, intelligent question answering system and chat robot appear constantly, but the semantics of these robots are built on the basis of the analysis of lexical relations, so it incapable to use the language to describe the image, human-computer interaction relates to the actual situation. The main problem lies in the language not associated with the actual perception.This paper designs a face image semantic representation parsing system based on the ARM and cloud calculation, the system can be used for intelligent equipment in human-computer interaction based on the face of semantic analysis on interactive objects, such as gender, age, emotion, appearance, and associated the language and face actual perception.The parsing process including three modules:face detection, feature extraction, and pattern classification. The face detection module use Haar-alike feature and Adaboost based method. Histogram of Oriented Gradient (HOG) and Local Binary Patterns (LBP) are employed for feature extraction in the subsequent face image. Finally, we adopted Linear Discriminant Analysis (LDA) for classification and give out the result.The system comprises a ARM terminal and cloud computing server. The terminal based on ARM embedded system, function with face detection and local, remote semantic representation parsing. The whole, hardware, system software, application software design were gave out, such as the design idea, flow chart and the main code, including the selection of ARM development board, the camera hardware interface, configuration of embedded Linux kernel, the embedded of OpenCV, the main code design etc, finally realized the terminal function.Cloud computing server use the aliyun Elastic Compute Service and cloud storage, provide face image semantic representation parsing analysis and model file upgrade service. It do work such as semantic representation model training, system management, representation database maintenance. This paper design that the whole structure, cloud computing, software, database, build a system management WEB page, and a visual configuration and management page for semantic representation model, gives the design of all data tables and fields in the representation database, finally provide parsing and upgrade service for the terminal.
|