Research On Face Detection And Recognition Algorithm Based On Image Processing

At present,science and technology are advancing by leaps and bounds,and the field of AI has attracted much attention.Face detection and recognition technology has become the focus of attention.there are mainly problems such as low detection rate and slow system detection rate in face detection technology,the accuracy and robustness of face detection technology still need to be improved.The main research content of this paper includes image preprocessing of face detection targets based on digital image technology and extraction of face image features based on improved Haar features.Then,the face detection of static face targets can be realized by combining the improved Haar features with Adaboost face detection algorithm.Then research on face target recognition in video,This paper combines Dlib with Open CV computer open source library with hardware and software equipment,extracts 68 key feature points of face target image,and calculates its feature value,then the actual detected face target image is compared with the face image in the face recognition database,and the face recognition system is finally designed.The specific contents of this paper are as follows:1.Image preprocessing and face feature extraction.Face image preprocessing includes face image normalization,face image grayscale,face image filtering and face image histogram equalization.Face image preprocessing can ensure the accuracy of face detection.Then,in the process of face feature extraction,considering the influence of face pose change on detection,the Haar feature is improved to further improve face detection rate and face detection robustness.2.Face detection is performed on the collected images.In this paper,the Adaboost algorithm is improved.Firstly,in the selection of weak classifiers,the quality of weak classifiers is improved by improving the selection method of weak classifiers,then,a strong classifier is constructed by a weak classifier.the computational complexity of the system is reduced and the detection efficiency of the system is improved to the greatest extent by cascading the strong classifier and combining with the improved Haar feature.3.The design of Face Recognition System.Based on the combination of Dlib and Open CV computer open source library,this paper designs the sub-modules of the face recognition system,uses the computer camera for video capture,and performs real-time face detection during the video capture process.The 68 key feature points of the face image are extracted,and the feature values are calculated.Then,the face recognition process is completed by comparing the features with the face database face image.