Face Sketch Synthesis via Sparse Representation-Based

Greedy Search
ABSTRACT:
Face sketch synthesis has wide applications in digital entertainment and law
enforcement. Although there is much research on face sketch synthesis, most
existing algorithms cannot handle some nonfacial factors, such as hair style,
hairpins, and glasses if these factors are excluded in the training set. In addition,
previous methods only work on well controlled conditions and fail on images with
different backgrounds and sizes as the training set. To this end, this paper presents
a novel method that combines both the similarity between different image patches
and prior knowledge to synthesize face sketches. Given training photo-sketch
pairs, the proposed method learns a photo patch feature dictionary from the
training photo patches and replaces the photo patches with their sparse coefficients
during the searching process. For a test photo patch, we first obtain its sparse
coefficient via the learnt dictionary and then search its nearest neighbors (candidate
patches) in the whole training photo patches with sparse coefficients. After
purifying the nearest neighbors with prior knowledge, the final sketch
corresponding to the test photo can be obtained by Bayesian inference. The
contributions of this paper are as follows: 1) we relax the nearest neighbor search
area from local region to the whole image without too much time consuming and

2) our method can produce non-facial factors that are not contained in the training
set and is robust against image backgrounds and can even ignore the alignment and
image size aspects of test photos. Our experimental results show that the proposed
method outperforms several state of-the-arts in terms of perceptual and objective
metrics.

EXISTING SYSTEM:
In Existing methods about face sketch synthesis could be sorted into three
categories: the subspace learning framework, the sparse representation based
approaches and the Bayesian inference framework.
Tang and Wang proposed principle component analysis based methods to
face sketch synthesis. These methods assumed that the mapping between a
photo and its corresponding sketch was a linear transformation. However,
due to the complexity of human face image, the relationship between face
photos and face sketches may preferably be estimated as a nonlinear
function.
Liu et al. adopted the idea of a locally linear embedding to model the
nonlinear process of face sketch synthesis. It is obvious that the LLE-based
method is insufficient to simulate the real nonlinear relationship between
photos and sketches. Further, these methods above need a great many of
training samples.

DISADVANTAGES OF EXISTING SYSTEM:

Existing algorithms cannot handle some non-facial factors, such as hair
style, hairpins, and glasses if these factors are excluded in the training set.
In addition, previous methods only work on well controlled conditions and
fail on images with different backgrounds and sizes as the training set.

PROPOSED SYSTEM:
In this paper, we developed a novel approach to face sketch synthesis by
incorporating both the similarity between different image patches and prior
knowledge. Greedy search based on sparse coefficients is adopted to
measure the similarity between the test photo patches and the training photo
patches. Intensity and gradient priors are employed to compensate the
greedy search stage.
Instead of directly employing raw test photo patches to search for nearest
photo patches in the training set, which is time consuming and requires huge
computational memory, we adopt sparse coefficients to replace the raw
image patches to overcome the aforementioned limitations. Moreover, by
sparse coefficients, we can expand the search range into the whole image,
which is impractical for existing patch level based methods due to the
computational complexity.

 In our method, the faces to be synthesized could possess some non-facial

factors, such as glasses and mustache etc.. The test photo can also be in
diverse poses with different backgrounds and sizes. The proposed method
can even deal with images including multiple faces.
ADVANTAGES OF PROPOSED SYSTEM:
We relaxed the search range from local area to the whole image via sparse
coding without increasing the computational cost too much.
The proposed face sketch synthesis method could handle some non-facial
factors, such as hair style, hairpins and glasses excluded in the training set
and different kinds of test photos ignoring image backgrounds, image size
and face posture etc..

SYSTEM ARCHITECTURE:

SYSTEM REQUIREMENTS:
HARDWARE REQUIREMENTS:
System

Pentium IV 2.4 GHz.

Hard Disk

40 GB.

Floppy Drive

1.44 Mb.

Monitor

15 VGA Colour.

Mouse

Logitech.

 Ram

512 Mb.

SOFTWARE REQUIREMENTS:
Operating system :

Windows XP/7.

Coding Language :

MATLAB

Tool

MATLAB R2013A

REFERENCE:
Shengchuan Zhang, Xinbo Gao, Senior Member, IEEE, Nannan Wang, Jie Li, and
Mingjin Zhang, Face Sketch Synthesis via Sparse Representation-Based Greedy
Search, IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 24, NO.
8, AUGUST 2015.