Font Size: a A A

Face Detection Study Based On Information Fusion

Posted on:2007-11-18Degree:MasterType:Thesis
Country:ChinaCandidate:Y LiuFull Text:PDF
GTID:2178360185494454Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
The techniques of face detection and identification have a bright future in the field of police (such as crime identification), safety validation, bank, customs, video conference, communication, digital amusement, floating population management and statistics. Harmless, unrestricted and credible face detection and identification are so required in many important occasions such as the system of ensuring public security, credit card affirmation, criminal identification, customs transit inspection and so on that it becomes researching hotspot in the field of contemporary model recognition and artificial intelligence. It has accumulated a great deal of fruits by years of researching on the techniques of face detection and identification. However, it is still faced with many difficulties under the non-restriction occasions. It is required to detect and segment human faces accurately, to have diversification compensation efficiently, to give character expression and accurately classification and to limit harm on receptor and to reduce the limitation on human liberty.In this paper we expound complexion Segmentation of color image, sub-image information amalgamation, and sub-image amalgamation human face detection. Combine with characters of human face this paper brings forward detection arithmetic of excessive scales image edge amalgamation. Two models of different colors (HIS and YCbCr) are applied to gain complexion image which has wavelet diversity then to get wavelet sub-image. Diversity of orthogonal wavelet Can not only save original information of figure image but also have the function to decompose. Daubechies1 orthogonal wavelet takes disperse wavelet diversity in the region of human face and Get horizontal sub-image LH, vertical sub-image HL, and crossing sub-image HH. Horizontal sub-image LH uses lines to detect diversity of rows. LH indicates horizontal edge of complexion region. Vertical sub-image HL indicates horizontal high frequency number of complexion region, vertical edge of the image. Sub-image on the cross HH show crossing edge of the image. In the...
Keywords/Search Tags:face detection, information fusion, wavelet transformation, face organ feature
PDF Full Text Request
Related items