Comparison between Thermal and Visible Facial Features on a Verification Approach - Biomedical Engineering Systems and Technologies - page 277

vocabulary tree, through the use of the k-means function. For each test image, only its
new descriptors are calculated and used to search through the hierarchical tree in or-
der to build a vote matrix, in which the most similar image of the database can be
easily identified. This approach mixes the singularity of the SIFT descriptors to per-
form reliable matching between different views of a face and the efficiency of the
vocabulary tree for building a high discriminative vocabulary. A description of the
system is provided in the next subsections.

2.1

Approach Proposed

This approach is composed by four stages: face segmentation, SIFT descriptors calcu-
lator, vocabulary tree construction and matching module.
While face segmentation is executed manually, the matching module searches in
the vocabulary tree the best correspondence between the test descriptors and those of
the database. Therefore, firstly the explanation is focused on the SIFT parameters and
tree classification, and secondly a brief description of the matching module is given.
A block diagram of the system is shown in figure 2.

SIFT descriptors
calculator

Vocabulary tree
construction

Visible and
Thermal image
database

Matching module

Verification

SIFT descriptors
calculator

Test Image Face segmentation

Fig. 2. Diagram of the proposed thermal face recognition system

2.2 Features Based on Scale-Invariant Feature Transform

The use of SIFT descriptors is applied in the most part of the results achieved by D.
Lowe in [15] as a guideline, only determinant parameters are modified in order to
adapt the algorithm to the system. Keypoints are detected using a cascade filtering,
searching for stable features across all possible scales. The scale space of an image,
L(x,y,σ) is produced from the convolution of a variable-scale Gaussian, G(x,y,σ) with
an input image, I(x,y);

L ( x , y , σ ) = G ( x , y , σ ) ∗ I ( x , y ) (1)

where * is the convolution operation in x and y, and

2 2

− ( x + y )

1

(2)

G ( x , y , σ ) = ⋅ e

2

2 σ

2 πσ

2