Open AccessLocal Directional Threshold based Binary Patterns for Facial Expression Recognition and Analysis
Author(s) -
V. Uma Maheswari,
Vara Prasad,
S. Viswanadha Raju
Publication year - 2018
Publication title -
international journal of engineering and technology
Language(s) - English
Resource type - Journals
ISSN - 2227-524X
DOI - 10.14419/ijet.v7i4.6.20225
Subject(s) - pixel , local binary patterns , artificial intelligence , pattern recognition (psychology) , computer science , standard deviation , precision and recall , co occurrence matrix , support vector machine , computer vision , binary number , mathematics , image (mathematics) , histogram , image texture , image processing , statistics , arithmetic
In this paper, proposing a novel method to retrieve the edge and texture information from facial images named local directional standard matrix (LDSM) and local dynamic threshold based binary pattern (LDTBP). LBP and LTP operators are used for texture extraction of an image by finding difference between center and surrounding pixels but they failed to detect edges and large intensity variations. Thus addressed such problems in proposed method firstly, calculated the LDSM matrix with standard deviation of horizontal and vertical pixels of each pixel. Therefore, values are encoded based on the dynamic threshold which is calculated from median of LDSM values of each pixel called LDTBP. In experiments used LFW facial expression dataset so used SVM classifier to classify the images and retrieved relevant images then measured in terms of average precision and average recall.