Font Size: a A A

Face Gender Recognition Technology Based On Digital Image Processing

Posted on:2015-04-03Degree:MasterType:Thesis
Country:ChinaCandidate:C LeiFull Text:PDF
GTID:2298330467453664Subject:Software engineering
Abstract/Summary:PDF Full Text Request
This article mainly researches the implementation process of facial genderidentification based on Matlab. In the pattern recognition field, it is alwaysconsidered to be a very challengeable study to identify people’s face utilizingcomputer science. It can also imposes good effect on man-machinecommunication and information acquisition, and will be a hot study orientationFace identification is a biometric identification technology that can implementpersonal identification based on people’s facial feature information. Also calledimage identification or figure identification, it is a series related technology thatutilize a camcorder or a camera to collect images or video stream containingpeople’s faces, and automatically detect and track human faces in the image, thento identify them.Face recognition is mainly used for identification. Because of the fast widespreadof video monitoring, many of the video monitoring application urgently need arapid identification technology, that can applied in a long distance and under theusers’ non-cooperation condition, so as to identify personal identity immediatelyover a long distance, realize the intelligent early warning. Face recognitiontechnology is undoubtedly the best choice, the fast face detection technology cansearch for a face in the surveillance video images in real time. And with real-timecomparison with face database, it can realize rapid identification.Face image preprocessing is the basis of face detection and recognition. Undernormal circumstances, all kinds of objective conditions that can interference and restrict the original image acquisition, such as the position o f the light source, theangle of the imaging, are uncertain. In addition, imaging tools, storagetransmission system and so on, will bring the unnecessary noise to theimage. Affected by these factors, the image data we get have different levels ofreduction of quality, or called degradation, such as low contrast, image edge blur,etc.Generally speaking, the face recognition system includes image intake, facepositioning, image preprocessing, as well as face recognition (identityrecognition or search).System input is usually one or a series of the face imagecontaining uncertain identity, and several human face image with knownidentity in the database, or the corresponding code. And its output is a series ofsimilarity score, suggests to the identity of the face yet to be identify.This article, by using matlab, so called the fourth generation of computer language,introduces the analysis process in detail how to do image preprocessing, facedetection and gender classification in the very environment, and research andinnovate in these three aspects respectively:(1)First, grey the colorful image; remove noise using the image filter. Then do thegrayscale transform to the image; enhance the image by adopting the method ofhistogram equalization. There are many kinds of transformation form of thismethod. In this article, we adopt continuous mean quantization transform toenhance the grey image.(2)In image face detection and localization of research, there are many featureextraction methods, because of the differences of applicable scope and difficulty,these methods can not be applied to the study of this article. This article mainlystudies the principal component analysis (PCA) algorithm, linear discriminateanalysis (LDA) and wavelet transform. According to the analysis and the testresults, face detection based on facefind function is proposed. (3)In the face gender classification method research, first extract face gendercharacteristics. In this article, we locate the human eye using gray integralprojection and then to extract gender related characteristics of nose, mouthrespectively.
Keywords/Search Tags:human face recognition, gender recognition, feature extraction, neuralnetwork, support vector machineFace recognition, Gender classification, Feature extraction, ANN, SVM
PDF Full Text Request
Related items