You are on page 1of 6

B.CHAITANYA KRISHNA* et al.

[IJESAT] INTERNATIONAL JOURNAL OF ENGINEERING SCIENCE & ADVANCED TECHNOLOGY

ISSN: 22503676
Volume - 2, Special Issue - 1, 93 98

AN EFFICIENT FACE RECOGNITION SYSTEM BY DECLINING REJECTION RATE USING PCA


B. Chaitanya Krishna1, V. H.Bindu2, K. L. Durga3, G.Lokeshwar4, G.AshokKumar5
Assistant Professor, CSE Department, K.L.C.E, A.P, India, chaitu2502@gmail.com 2 Student, CSE Department, K.L.CE, A.P, India, himabindu.valluri@gmail.com 3 Student, CSE Department, K.L.CE, A.P, India, lakshmidurgakonidena@gmail.com 4 Student, CSE Department, K.L.CE, A.P, India,gallalokeswar@gmail.com 5 Student, CSE Department, K.L.CE, A.P, India, g.ashokkumar125@gmail.com
1

Abstract
Face detection is an important step in face recognition which is one of the more representative and classic application in computer vision. Human face recognition has gained wide attention as a covert method for surveillance to enhance security in a variety of application domains. Face is one of the physiological biometrics based on stable features and authenticates an individual uniquely. Face detection is a challenging mission and is user friendly. Here we use PCA (Principal Component Analysis) which belongs to the appearance based approach. The given image is going to be experienced with Pre-processing section include several factors like gray scale conversion ,contrasting , image resizing, Histogram equalization in order to enhance the quality and to make time complexity, rejection rate as low for the effective face recognition.

Index Terms--- Face Recognition, PCA, Eigen Vector, Eigen Value, Rejection rate, CDF, RGB model. ---------------------------------------------------------------------***------------------------------------------------------------------------1. INTRODUCTION
Computer Vision is a very new area of research is currently

hanging in the Computer Science under. It mainly wants to solve the problem by building an intelligent system. The ultimate purpose of computer vision is to extract the useful information from the images (Image Processing) and to create such as like humans can be made for the image the wisdom of response and identification systems. Face Recognition is one of the main applications. It includes the identification and verification of the people. Face recognition provides passive identification that is the person to be identified does not need to cooperate or take any specific action. For example, a smart store can recognize its regular customers while they are entering the store. The customers do not need to talk or look directly to the camera to be recognized. The main problem in machine recognition is that the human faces are non-rigid objects and appeared in different scale, pose, angle and facial expressions. Human faces always have variations for example, glasses. In additional, the images have

different brightness. For any type of image including the above factors the output should come in such a way that the time complexity and the rejection rate should be small. The machine recognition has limited memory to recall a person immediately and also the stored database is limited. This must not be the case with the machine recognition system where the memory is manifold. In this paper we mainly focus on several preprocessing factors such as gray scale conversion, Image resizing, contrasting and histogram equalization. Prior to recognition, face detection will be done. Face Detection is a technology to determine human face in an image. It just only concerns facial features in testing images. The detected faces are preprocessed in order to increase the efficiency of the face recognition system. Face recognition system having two applications. It either recognizes only one image (computer user authentication) or many images in case of family members identification. In both of the cases the image correspond to the input are going to be recognized. The no of rejected images in the database is going to be calculated and finally rejection rate is calculated.

IJESAT | Jan-Feb 2012


Available online @ http://www.ijesat.org 93

B.CHAITANYA KRISHNA* et al. [IJESAT] INTERNATIONAL JOURNAL OF ENGINEERING SCIENCE & ADVANCED TECHNOLOGY

ISSN: 22503676
Volume - 2, Special Issue - 1, 93 98

2. APPROACHES TO FACE RECOGNITION SYSTEM


The main methodology include a. Pre-processing: This stage includes images processing to enhance the quality of input images. b. Face detector stage: The core section in face detection includes PCA. Face detection algorithms mainly focused to detect the frontal human face in early days. At present, the techniques are proposed focused on the face recognition and increase the need of face detection. These are helpful in providing security and authentication. There are mainly four approaches to detect the face.

Cons: Not suitable to detect facial features from uncontrolled background, time consuming algorithms. Detection rate is not accuracy, because of need to combine different feature and processing it.

2.4. Learning Based Method/Appearance Based


Method: The models are trained from a set of training set before doing detection. For the large amount of training data, it can be provided high accuracy recognition rate to resist variation, expression and pose of faces images. For instance, many of non-face and face images import into the system. Machine learning techniques are employed to train the system based on the statistical properties and probability distribution function. Principle Component Analysis (PCA) is an example for this type of approach.

2.1. Knowledge based methods: The models are used


human knowledge to find a face patterns from the testing images. Based on the nature of human faces, algorithms scan the image from top-to-bottom and left-to-right order to find facial feature. For instance, face should be including two eyes and mouth etc. Pros: Easy applied into simple rules. Cons: difficult to detect in invariant background, such as different pose, uncontrolled illumination. This algorithm does not work on the pose.

3. PREPROCESSING FACTORS 3.1. Grayscale Conversion


In order to reduce the information of images, image should be done a converting to grayscale. Each color images (RGB images) are composed of 3 channels to present red, green and blue components in RGB space. Image is defined with grayscale level, which means pixels in images are stored 8-bit integer to represent color from black to white. In addition, grayscale images are sufficient enough to process face detection. Various algorithms include are 1. The lightness method averages the most prominent and least prominent colors: (max(R, G, B) + min(R, G, B)) / 2. 2. The average method simply averages the values: (R + G + B) / 3. 3. The luminosity method is a more sophisticated version of the average method. It also averages the values, but it forms a weighted average to account for human perception. Were more sensitive to green than other colors, so green is weighted most heavily. The formula for luminosity is 0.21 R + 0.71 G + 0.07 B.

2.2. Template marching: The model is used several


templates to find out the face class and extract facial features. Rules are pre-defined and decide whether there is face in the image. By using filters also we can extract the contours of face shape. Pros: Simple to apply this method. Cons: similar to knowledge based method, hard to detect face in different poses. Algorithms are sensitive to scale size, face shape and pose

2.3. Invariant feature methods: The model is bottom-up


approaches and used to find a facial feature (eyebrows, nose), even in the presence of composition, perspective vary, it is difficult to find a face real time using this method. Statistical models are developed to determine the faces. Pros: Unlike knowledge-based method, it is invariant to pose and expression.

IJESAT | Jan-Feb 2012


Available online @ http://www.ijesat.org 94

B.CHAITANYA KRISHNA* et al. [IJESAT] INTERNATIONAL JOURNAL OF ENGINEERING SCIENCE & ADVANCED TECHNOLOGY

ISSN: 22503676
Volume - 2, Special Issue - 1, 93 98

3.2. Contrasting:
It means to make image brighter for the visualization purpose. 1.The algorithm it uses for contrast enhancement takes the red, green, and blue values separately and boosts them by 1.2 times if they are already brighter than 128. r= r*1.2,g=g*1.2,b=b*1.2. If they are below 128, then they are divided by 1.2. r= r/1.2,g=g/1.2,b=b/1.2 2. Color brightness is determined by the following formula is ((Red value X 299) + (Green value X 587) + (Blue value X 114)) / 1000

In order to map the values back into their original range, the following simple transformation needs to be applied on the result: y' = y. (max{ x } min{ x} ) + min {x}The general histogram equalization formula is

Where cdfmin is the minimum value of the cumulative distribution function, M N gives the image's number of pixels, L is the number of grey levels used.

3.3. Histogram Equalization


Consider a discrete gray scale {x} and let ni be the number of occurrences of gray level i. The probability of an occurrence of a pixel of level i in the image is Px(i)= p(x=i) = (ni / n) , 0 i <L L being the total number of gray levels in the image, n being the total number of pixels in the image, and P x(i) being in fact the image's histogram for pixel value i, normalized to [0,1].The cumulative distribution function corresponding to px (the image's accumulated normalized histogram) is cdfx(i)= x(j) We would like to create a transformation of the form y = T(x) to produce a new image {y}, such that its CDF will be linearized across the value range, i.e. cdfy(i)=iK for some constant K. y=T(x) = cdfx(x) .

3.4. Image Resizing:


Images are synthesized by numerous of pixel which is the small unit in the image. Also, images are the 2-dimensional matrix pattern, every pixel in the images is represented something information. For example, 0 is white and 255 is black in gray scale images. Because there are a lot of information to deal with, input images are employed the resizing processing to reduce the images resolution with keeping same quantity. If the image size is either large or small we cannot see the image clearly. Below example is an illustration about the different resolution to describe the same image

IJESAT | Jan-Feb 2012


Available online @ http://www.ijesat.org 95

B.CHAITANYA KRISHNA* et al. [IJESAT] INTERNATIONAL JOURNAL OF ENGINEERING SCIENCE & ADVANCED TECHNOLOGY

ISSN: 22503676
Volume - 2, Special Issue - 1, 93 98

4. PCA
It is probably the most widely-used and well-known of the standard multivariate methods and invented by Pearson (1901) and Hotelling (1933) , first applied in ecology by Goodall (1954) under the name factor analysis (principal factor analysis is a synonym of PCA").It belongs to lossy operation in dimensionality reduction approach. Problems arise when performing recognition in a high-dimensional space. Significant improvements can be achieved by first

mapping the data into a lower-dimensional sub-space. Dimensionality reduction implies information loss.PCA is a technique that can be used to simplify a dataset and it is a transformation that chooses a new coordinate system for the data set such that the greatest variance by any projection of the data set lies on the first axis (1st principal component), the second greatest variance lies on the second axis (2nd principal component), and so on. The goal of PCA is to reduce the dimensionality of the data while retaining as much as possible of the variation in the original dataset.

PCA ARCHITECTURE

IJESAT | Jan-Feb 2012


Available online @ http://www.ijesat.org 96

B.CHAITANYA KRISHNA* et al. [IJESAT] INTERNATIONAL JOURNAL OF ENGINEERING SCIENCE & ADVANCED TECHNOLOGY

ISSN: 22503676
Volume - 2, Special Issue - 1, 93 98

4.1. Training the system:


1. Acquire the set of face images known as the training set. Basically the images in the training set are of size NxN so we have to represent the original images in the form of vectors i.e., of the size N2x 1. Let us denote these vectors as 1, 2, 3,. p. Where p=no of images in the training set. Now find the average face of the above set by using the formula i. = (1/N)pi=1 i where is the mean image with dimensions N2x 1. The centered images are found by a. i= i- where i=1..p The Covariance matrix is constructed as C=A.AT where A = [1 , 2 .p] ,the dimensions of C is N2xN2 ,here we can observe that the dimensionality is very high so let us take the reverse case as C = AT.A so that dimensionality has been reduced to pxp. Now obtain the Eigenvectors and Eigen values for the covariance matrix. Let us consider V as Eigenvector and U as Eigen value CV= UV. Order the Eigen Vectors with respective to Eigen Values. Keep only the eigenvectors associated with non-zero Eigen values. This matrix of eigenvectors is the Eigen space V, where each column of V is an Eigenvector V =[v1 | v2 | ... | v q] where q<p. Project the centered training images into the Eigen Space; calculate the dot product of the image with each of ordered Eigenvectors. i=VT.i Therefore, the dot product of the image and the first Eigenvector will be the first value in the new vector. The new vector of the projected image will contain as many values as Eigenvectors.

3. Calculate the Euclidean distance by using formula 2i=||input -i||2 The image with the least Euclidean distance is considered as the closest image.

5. SOLUTION PROCESS
1. Select the trained data set folder having trained images for all images present in the test database.

2.

3.

4.

5.

6.

7.

2. Select the Test dataset folder having test images and then select the input image from the test dataset folder.

8.

4.2. Testing the system:


Here the input image is given to the face recognition system for recognition. 1. The input image is converted to a vector of size N2x1. This vector is considered as . 2. Project the input vector into the Eigen Space input=VT. input 3. The selected images in trained database are detected as follows.

IJESAT | Jan-Feb 2012


Available online @ http://www.ijesat.org 97

B.CHAITANYA KRISHNA* et al. [IJESAT] INTERNATIONAL JOURNAL OF ENGINEERING SCIENCE & ADVANCED TECHNOLOGY Journal of Global www.jgrcs.info.

ISSN: 22503676
Volume - 2, Special Issue - 1, 93 98 Research in Computer Science,

3. Computer vision - Wikipedia, the free encyclopedia, http://en.wikipedia.org/wiki/Computer vision 4. Evaluation of image pre-processing techniques for Eigen face based face recognition, The Proceedings of the Second International Conference on Image and Graphics, 2002 Thomas Heseltine, Nick Pears and Jim Austin 5. Contrasting-Java2 Complete Reference, Herbertschildt 6. PCA-Data mining Techniques-KAMBER 7. Gray Scale, http://www.johndcook.com/blog/2009/08/24/algorithmsconvert-color-grayscale/ 4. Now we have to find the rejection rate and rejection percentage. Rejection rate=1 Rejection Percentage=(1//9)*100=11.1% 5. By resizing the images we can reduce the rejection rate for a particular size of an image. 6. And the time complexity may become small as we apply preprocessing factors before applying the algorithm. 8. Histogram equalization, http://en.wikipedia.org/wiki/Histogram_equalization 9. Contrasting, http://juicystudio.com/article/luminositycontrastratioalgorithm .php 10. Face Recognition For Smart Interactions ,H.K. Ekenel, J. Stallkamp, H. Gao, M. Fischer, R. Stiefelhagen, interACT Research, Computer Science Department, Universitt Karlsruhe (TH) ,76131, Karlsruhe, Germany, e-mail: ekenel@ira.uka.de, web: http://isl.ira.uka.de/ 11. Face Recognition Using Eigenfaces,M. Sc. THESIS 1996,ILKER ATALAY,Eigenfaces_Msc_Thesis.Pdf 12. Face Detection, http://en.wikipedia.org/wiki/Face_detection 13. L.I. Smith. A tutorial on principal components Analysis, www.plantbiology.siu.edu/PLB444/PCA.ppt 14. Martinez, A. Kak, "PCA versus LDA", IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 23, no. 2, pp. 228-233

6. CONCLUSION
Face recognition is a biometric identification which identifies the individual uniquely.PCA for face recognition is fast and simple. Several problems involved with various poses, illumination, aging etc. of the images are eliminated by using the PCA. Image Resizing, Gray Scale Conversion, Contrasting, and Histogram equalization plays an important role in Pre-Processing. Because of these factors the image can be recognized with less effort. So the time complexity and rejection rate becomes small to keep the recognition rate high. And hence effective face recognition can be possible.

REFERENCES
1. Yales Face Database, http://cvc.yale.edu/projects/yaleface/yalefaces.html 2. Y. Vijaya lata and dr. A. Govardhan ,Impact Of Image Resizing Factor In Face Recognition System Using PCA,

IJESAT | Jan-Feb 2012


Available online @ http://www.ijesat.org 98

You might also like