Dematting face images for use in a face recognition system

Senior Thesis for obtaining a Master's degree

Abstract

Face recognition is in the fields of biometrics, machine vision and pattern recognition and has a wide application, including issues related to security systems. Since various factors such as ambient lighting, noise, and image opacity are more or less effective in the performance of face recognition methods, therefore, it is particularly important to examine the methods of removing blur from face images used in face recognition algorithms, in order to improve recognition accuracy.

Given this issue that the basic problem in all image enhancement methods is to understand the type and characteristics of the point spread function (PSF) related to the image blurring factor, therefore, in a part of the proposed method in This thesis, by learning previous knowledge from training examples that include artificially blurred face images, has been addressed to identify the PSF of the blurring agent of face images. According to the proposed method in this thesis, first in the training phase, we artificially matte the sets of facial images related to the ORL database using a few specific PSFs and then add white noise with an average power of 30 dB to them. Now, we put the features consisting of the maximum information related to the size of the frequency components of the matted images with the same PSF in one category and use the MLP neural network to learn the knowledge from the created feature space. Then, in the test phase, the input opaque face image, which has an unknown opaque PSF, is mapped to the feature space of the training phase, and we extract the previous features from the mapped image. Now, with the help of previously trained neural network, we select the category closest to this image from among the trained categories, and consider the blurring PSF of the images related to this category as the blurring PSF of the input matte face image. In the following, according to this PSF and using deconvolution method (convolution photo), we improve the input image and deliver the improved image to the face recognition system to perform the recognition process.

During the proposed method in this thesis, by creating a special feature space consisting of the maximum information related to the size of the frequency components of opaque images, we succeeded in increasing the PSF identification accuracy (identification accuracy higher than 80% in noisy conditions) and as a result, increasing the accuracy of the recognition system. face (increasing the accuracy of the recognition system from 19.833% to 90.837%) by this method. Also, the use of neural network to identify PSF, on the one hand, has reduced the average execution time of this method by 172.41% compared to the modern methods presented in this field, and on the other hand, it has increased the hardware implementation capability of this method compared to the existing methods.

Key words: removing blur from face images, face recognition systems, point expansion function, feature space learning, MLP neural network

1-1 Introduction

Subject Face recognition is widely used in fields such as image processing, machine vision, pattern recognition, neural networks and machine learning. The face recognition system is a biometric system (biometer) [1] that identifies or confirms the identity of a human using intelligent and automatic methods. Several methods have been presented for face recognition, and these methods are generally divided into the following two categories: A) Pattern-based methods Pattern-based methods operate based on the comparison of the input image with sets of patterns related to the structure of the face. These patterns are created using statistical tools such as support vector machine (SVM[2]), principal component analysis (PCA[3]), linear discriminant analysis (LDA[4]) and independent component analysis (ICA[5]) from face images related to the training set. In these methods, people's faces are matched with a predetermined model and the obtained data are stored as extracted features.

One of the advantages of the face recognition system is the appropriate power of identification, harmlessness, friendliness and naturalness of the method to recognize people. On the other hand, it should be mentioned that, unfortunately, automatic face recognition with the help of a machine still remains a scientific challenge.Among the reasons for this lack of success, we can mention the large amount of image data and, accordingly, the wide range of changes in this data, as well as the nature of the data itself. For example, the identification system should be less sensitive to age conditions and face makeup, hair style, facial expressions, facial angle changes, etc. be resistant To the mentioned factors, the imaged opacity, lighting conditions, the background of the images and other imaging parameters can also be added. 1-2 Statement of the problem In general, a face recognition system consists of the parts shown in figure (1-1).

According to the process of the view shown in Figure (1-1), the input data to a face recognition system includes a video or image of an unknown person along with other objects. To recognize these people by the desired recognition system, we need to separate the area related to the faces of the people. For this purpose, a facial recognition system is used. On the other hand, with the advancement of face recognition methods, image blurring factors will not have a significant effect on the performance of face recognition systems [1]. Therefore, as you can see in Figure (1-2), the area related to the person's face in three transparent images, the image with weak blurring, and the image with strong blurring, has been correctly recognized. In this way, we do not need to perform pre-processing to recognize the location of the face.

However, since the blurring of the images weakens the recognition accuracy, therefore removing the blurring from these images before the feature extraction and classification steps will greatly help to increase the recognition accuracy.

Figure (1-2): An example of face recognition in matte images using the method described in [1]

(a) Image with severe blur (b) Image with matte Weakness (c) transparent image

1-3 The necessity of conducting research and the purpose of the thesis

As it was said, face recognition systems work very poorly on images that are opaque. Therefore, improving such images helps a lot to increase their recognition accuracy. In this thesis, our goal is to design and simulate a method to improve blurred face images in order to improve the efficiency of face recognition systems. According to the review given in the next chapter for these methods, it can be seen that the number of available methods for improving the blurred face image in order to increase the efficiency of face recognition systems is very small. In chapter 3, we introduce the proposed method in this thesis. In chapter 4, the results of the simulation of the proposed method to identify the point spread function [6] (PSF) of matting input face images and de-matting them in order to be used in a face recognition system are reviewed. Chapter 5 is dedicated to expressing the conclusion and proposing the future solution.

As it was said, all environmental conditions such as the change of ambient light or blurring of the image, are effective in face recognition methods, therefore, investigating the effect of de-blurring methods of incoming images will be of particular importance in improving the recognition accuracy of face recognition algorithms. First, as seen in Figure (2-1), the characteristics of each person's image are seriously changed under the influence of image blurring factors; Secondly, the face images of different people become more similar when blurred (Figure (2-1)) and therefore the recognition of these images has a significant error.

Figure (2-1): The effect of blurring factors on face images [20]

Although these two problems greatly reduce the accuracy of face recognition systems, few methods have been presented to deal with these problems and none of them have given satisfactory results. They do not include the removal of opacity related to facial images, especially in the real world application, and the increase of recognition accuracy.

In this chapter, in order to learn more about the methods of removing opacity from images, first, in section 2-2, we will examine the new methods of removing opacity from general images. Then, in section 2-3, we will examine the methods of removing opacity from face images. 1-4 Methods of removing opacity from general images The issue of image opacity removal is widely discussed in areas such as image processing, computer graphics, and machine vision.