Sunteți pe pagina 1din 16

FACE RECOGNITION USING LAPLACIANFACES

ABSTRACT

Face recognition systems are computer-based security system that are able to automatically detect and identify human face. It is an appearance-based methods, we usually represent an image of size n x m pixels by a vector in an n x m dimensional space. It is an appearance-based face recognition method called the Laplaciance approach. It uses Locality Preserving Projections (LPP) the face images are mapped into a face subspace for analysis. Contd..

Continuation.. It is different from Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) which effectively see only the Euclidean structure of face space . LPP finds an embedding that preserves local information, and obtains a face subspace that best detects the essential face manifold structure. The Laplacian faces are the optimal linear approximations to the Eigen functions of the Laplace Beltrami operator on the face manifold. In this way, the unwanted variations resulting from changes in lighting(left-light,center-light,right-light), facial expression(Normal,Happy,Sad,Sleepy,and surprised), and pose(with / without glasses) may be eliminated or reduced.

Existing System

Principal Component Analysis (PCA) is an eigenvector method designed to model linear variation in High-Dimensional data. PCA can do are prediction, redundancy removal, feature extraction, data compression, etc. The main idea of using PCA for face recognition is to express the large 1-D vector of pixels constructed from 2-D face image into the compact principal components of the feature space. This is called Eigenspace projection.

PCA DISADVANTAGES Less accurate. Does not deal with manifold structure . It doest not deal with biometric characteristics.

PCA and LDA


PCA used in Linear Model such as Signal Processing. Image Processing. System & Control Theory and Communication etc. PCA AND LDA ( Will not consider LOCAL STRUCTURE) PCA reduces the dimensionality of image space by combining features. LDA projects the high Dimensional data onto a lower dimensional data.

Proposed System
Propose a new approach to face analysis . Each face image in the image space is mapped into a low dimensional face subspace, which is a characteristic by a set of future images called Laplacianfaces.

Manifold Structure.

The manifold structure is modeled by a nearest-neighbor graph which preserves the local structure of the image space. A face subspace is obtained by Locality Preserving Projections(LPP). Each face image in the image space is mapped into a low dimensional face subface.

How to use Laplacianface for Representation


PCA Projection. Construction of Nearest Neighbor Graph. Choosing the Weightage. EigenMap.

PCA PROJECTION We project the image set into the PCA subspace by throwing away the smallest principal components. CONSTRUCTION OF NEARST NEIGHBOR GRAPH The constructed nearest neighbor graph is an approximation of the local manifold structure.

CHOOSING THE WEIGHTAGE


If node i and j are connected, put 2 - || xi xj|| S =e __________ ij t Where t is a suitable constant. Put Sij = 0. The weightage matrix S of graph G. Model the face manifold structure by preserving local structure. EIGEN MAP

Compute the Eigen Vector & Eigen Values.

Face Manifold Analysis


The face image are divided into two parts. The left part include the face images with open mouth. The Right part of the images are with the closed mouth. It makes the neighboring points in the image face nearer in face space and faraway points in the image face farther in the face space. The bottom image corresponding to points along the right path (Linked by solid lines.)

Face Manifold Analysis

Two dimensional linear embedding of face images by Laplacianfaces. AS can be seen, the face images are divided into two parts, the faces with open mouth and closed mouth. Moreover, it can be clearly seen that the pose and expression of human change continuously and smoothly.

FACE RECOGNIZATION USING LAPLACIANFACES.


STEPS IN FACE RECOGNIZATION The Laplacianfaces from the training set of face images is calculated. The new face image to be identified is projected into the face subspace spanned by the Laplacianfaces Finally the new face image is identified by the nearest neighbor classifier.

Experiments

Original images were normalized (in scale & Orientation) such that the two eyes were aligned at the same position. Then the facial alias were cropped into the facial images for matching. The size of each cropped image in all the experiments in 32 x 32 pixel with 256 gray levels per pixels.Each image in represented by a 1024 dimensional vector in images space. Different pattern classifiers have been applied for face recognition, includes nearest-neighbor, Bayesian, etc..We apply the nearest neighbor classification for its simplicity.

Databases used in Laplacianface Recognition

YALE DATABASE It contains 165 gray scale images of 15 individuals. The images demonstrate variation in

light condition Facial Condition. With / without glass

PIE DATABASE

It contains 68 Subject with 41,368 face images, It has been taken by many cameras and various pose illumination and expression. This database was collected at Microsoft Research Asia. It contains 12 individuals, captured in two different sessions with different backgrounds and illuminations.

MSRA DATABASE

Results of Experiments
All the Three Approaches are are compared on each databases. The result came as LAPLACIANFACES can identify the person with different expressions, poses and lighting conditions than the other two.

CONCLUSION AND FUTURE WORK

The manifold ways of face analysis (Representation and recognition) are introduced in this paper in order to detect the underlying nonlinear manifold structure in the manner of linear subspace learning. The manifold structure is approximated by the adjacency graph computed from the data points. Using the Laplacian of the graph, I then computed a transformation matrix which maps the face images into a face subspace.

S-ar putea să vă placă și