Software Simulation And Hardware Implementation Of Face Detection Algorithm For Security Video Surveillanc

With the continuous development of society,people have higher and higher requirements for security monitoring for the safety of their own property and life,and the most important thing in security monitoring is the detection of human faces.Since the face is a dynamic object,its appearance is highly variable.For the face itself,face detection at different angles,different poses,and different expressions will bring great difficulties to detection.In terms of the external environment,when the light is not strong,such as when the light is dim,or at night,or the color of the background environment is similar to that of a human face,or there are obstacles blocking the detection environment,the detection environment will also be accurate.bring great difficulties.Therefore,how to accurately detect faces is an urgent problem that needs to be solved at present.In this regard,in view of various problems encountered in the process of face detection,this paper proposes an algorithm based on improved skin color space joint Adaboost face detection.First,the research background and significance of this paper,and the research status at home and abroad are expounded.On this basis,a face detection algorithm based on skin color is proposed,which uses RGB norm to improve Gamma correction for the normalization problem of Gamma correction;creates a new color space YCbCgCr,establish a threshold skin color model to perform preliminary detection on the face,and obtain candidate images;construct a Haar feature of "工" type that is more in line with the face features,improve the weight of Adaboost samples,and detect candidates through the cascade classifier constructed by Haar features.face in the area.After experimental research on various face images,it is found that the method has higher accuracy than the existing Viola–Jones technique,and can detect side or occluded faces,especially under different expressions.Faces can be detected accurately.Secondly,the software simulation experiment is carried out on face detection,The GUI graphical interface is established through Matlab,and the face detection algorithm proposed in this paper is compared with the traditional Adaboost face detection algorithm.The results show that the face detection algorithm proposed in this paper is The accuracy of the detection algorithm has been improved,and the false detection rate has also been greatly reduced,of which the accuracy rate is as high as 92%.Finally,port the algorithm to Altera’s DE1＿SoC development board.Conducted system process design,construction of Linux development environment,cross-compilation of Qt and OpenCV libraries,design of FPGA acceleration module,and hardware experiments of algorithms.Faces are detected,and the results show that the detection accuracy and processing time are further improved.