Chapter 12
Landmark localization for face
recognition
12.1 Introduction . . . . . . . . .. . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . .. . . . . . 253
12.2 Elastic bunch graph matching . . . . . . .. . . . . .. . . . .. . . . . .. . . . .. . . . . . 253
12.3 Gabor wavelets . . . . . . . .. . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . . .. . . 254
12.4 Gabor jets . . . . . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. . 256
12.5 The elastic bunch graph matching algorithm . . . . .. . . . .. . . . . .. . . . 258
12.6 Application to face recognition . . . . . . . . .. . . . . .. . . . .. . . . . .. . . . . .. . 260
12.7 Facial landmark detection . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . .. . . . . .. 261
12.7.1 ASEF correlation filter . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. 261
12.7.2 Formulation of ASEF . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . .. . . 262
12.8 Eye detection . . . . . .. . . . .. . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . . .. . 263
12.9 Multicorrelation approach . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . . .. . . 264
12.9.1 Design of landmark filter(LF) . . . . . . . . .. . . . .. . . . . .. . . . . .. 264
12.9.2 Landmark localization with localization filter . . . .. . . . . .. 267
12.10 Test results . .. . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. . 268
12.1 Introduction
Detection of facial landmarks is one of the first steps for face detection
and recognition, emotion recognition, alignment of face images and many
more [112]. The elastic bunch graph matching (EBGM) method provides a
promising facial landmark detection where Gabor-jets are used to evaluate
the features of landmarks and a face graph is developed to automatically
select the test landmarks. In this chapter both EBGM and average synthetic
exact filter applications towards landmark localization are given.
12.2 Elastic bunch graph matching
Elastic bunch graph matching (EBGM) is a feature-based face
identification method. The algorithm assumes that the positions of certain
253

Get Face Detection and Recognition now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.