Difference of Gaussian Oriented Gradient Histogram for Face Sketch to Photo Matching

Retrieving a corresponding photo from a forensic sketch remains a great challenge. One of the difficulties is that the retrieval rate reduces significantly on data sets with shape exaggeration and lighting variation. The goal of this paper is twofold. First, we attempt to reduce the factor of the sh...

Full description

Saved in:
Bibliographic Details
Published in:IEEE access 2018-01, Vol.6, p.39344-39352
Main Authors: Setumin, Samsul, Suandi, Shahrel Azmin
Format: Article
Language:English
Subjects:
Alignment
Difference of Gaussian
Distance measurement
Dogs
Face
Feature extraction
Histograms
identity of interest
Illumination
image matching
Lighting
Matching
Mouth
oriented gradient histogram
Shape
sketch to photo
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Retrieving a corresponding photo from a forensic sketch remains a great challenge. One of the difficulties is that the retrieval rate reduces significantly on data sets with shape exaggeration and lighting variation. The goal of this paper is twofold. First, we attempt to reduce the factor of the shape exaggeration problem by introducing new fiducial points for geometric face alignment. Next, to minimize the illumination effects, we describe the image using the difference of Gaussian-oriented gradient histogram. The matching results using the simplest distance measure on two public databases indicate that our proposed face sketch to photo matching method achieves significantly better accuracy than the state-of-the-art methods. It is also proved that shape exaggeration influences can be reduced by employing outer points face alignment.
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2018.2855208