FRUE - Face Recognition Using Eigens
Principal Component Analysis | Fisher Discriminant Analysis |
|
A face recognition method based on PCA (Principal Component Analysis) and LDA (Linear Discriminant Analysis) is implemented. The method consists of two steps: First, we project the face image from the original vector space to a face subspace via PCA, and then we use LDA to obtain a linear classifier by means of which the input image given is suitably classified. PCA is often used for projecting an image into a lower dimensional space or face space and then LDA is performed to maximize the discriminatory power. The face recognition system begins by collecting a large number of facial images in a database. The system then creates a set of Eigen faces by combining all of the facial images in the data base and comparing the commonalities and differences between groups of individual facial images. The basic idea of this system is to obtain the relevant information for classification by using the Eigen faces in the database. |
|
|
Face Recognition Project is not in development now.
The Face Recognition project is not in development now. If you want to take a look at the source code, go to the project Home page. Special Thanks to OpenCV for Image Processing. Front End is developed using the .Net Framework. |
Principal Component Analysis
We form our database by adding 5 or more images for each person. Pre Processing of each image is done; this includes normalizing the shifts and orientations of the head position, histogram equalization and background removal. PCA (Principal Component Analysis) Then we form the Eigen faces for each person. The Eigen faces are formed by decomposing the face images into a small set of characteristic feature images. Each individual is characterized by a small set of weights needed to describe and reconstruct them. Each individual face can be represented exactly in terms of a linear combination of the eigenfaces The weight vector for the new input image is constructed with the help of the eigenfaces that were already stored in the database. The weight of the new input image is then compared with the weight vector of every face library member within a threshold. If there exists at least one face library member that is similar to the acquired image within that threshold then, the face image is classified as "known". Else a miss has occurred and the face image is classified as "unknown". The key aspect of this system is that the most relevant features or the principal components are extracted for recognition. Even if some features of the face are obscured, the system is still able to perform perfect recognition as it extracts the most relevant features necessary for recognition. Hence this method is known as Principal Component Analysis The advantages of this method over the feature based method are its speed and simplicity.
Fisher Discriminant Analysis
References
|
|
A project by Parameswaran M. |