Font Size: a A A

Research On Face Recognition Based On Optimized B-LBP

Posted on:2013-01-26Degree:MasterType:Thesis
Country:ChinaCandidate:X H ZhaoFull Text:PDF
GTID:2218330374965291Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
In the broad sense, an automatic face recognition system is consist of four important parts:face detection, facial feature points location, facial feature normalization, face recognition. Face detection finds if there are any faces in the image or video sequences, and acquires the location and extent of each face. Facial feature points should be located and normalization should be done after the first two steps. Face recognition identifies the person in the scene using a stored face database in the last.The paper mainly work on:proposed the optimized B-LBP pattern, optimized an eye locate method based on gray value projection peak analysis and optimized scaling algorithm.B-LBP pattern was born in order to make LBP pattern provide more information. B-LBP can provide more facial feature information, but some redundancy information were included all the same. To solve this problem, this paper raise optimized B-LBP, which may find the effective area and abandon the areas which were more easily affected by expressions. This method can improve the effective dimension while keep dimension lowThe optimized gray value projection peak analysis algorithm for eye location fully used the face law, which can locate the eyes in the face efficiently, focusing on the point that human eyes have the highest gray value than other points, an optimized algorithm is proposed to make this locate algorithm more robust.Normalization is the important step of face recognition, in this paper, a optimized scaling algorithm is applied. This method can keep the original shape of face and guarantee the accuracy.
Keywords/Search Tags:Local Binary Pattern, face detection, face recognition, eyes location, face normalization
PDF Full Text Request
Related items