26-05-2012, 03:20 PM
Application of DCT Blocks with Principal Component Analysis for Face Recognition
[attachment=23184]
Introduction
Over the past few years, the user authentication is increasingly important as the security control is required everywhere. Traditionally, ID cards and passwords are popular for authentication although the security is not so reliable and convenient. Recently, biological authentication technologies (i.e Biometrics) through voice, fingerprints, iris print, retina, palm print, face..etc is playing an important role in modern personal identification systems. The face, among them, is chosen for the suggested system. The face is user friendly as people are accustomed to taking photographs. It is economic due to the low cost of cameras and computers (it does not require any special devices).
The Discrete Cosine Transform
The DCT is a popular technique in imaging and video compression, which was first applied in image compression in 1974 by Ahmed et al. In 1992, the first international standard for image compression, known as the Joint Photographic Experts Group (JPEG), was established with the DCT encoder and decoder.
The Principal Component Analysis
Eigenface is one of the most thoroughly investigated approaches to face recognition. It is also known as Karhunen-Leove expansion, eigenpicture, eigenvector, and principal component [12]. Sirovich and Kirby and Kirby et al. used PCA to efficiently represent pictures of faces. They argued that any face image can be approximately reconstructed by a small collection of weights for each face and a standard face image (eigenpicture). The weights describing each face are obtained by projecting the face image onto the eigenpicture. In mathematical terms, eigenfaces are the principal components of the distribution of faces, or the eigenvectors of the covariance matrix of the set of face images.
Recognition Results
The used data set is from the Olivetti Oracle Research Lab (The ORL database found in [14] ). The ORL database consists of 400 frontal faces; 10 tightly cropped images of 40 individuals with variations in pose, illumination, facial expressions and accessories. The size of each image is 92×112. The training set is composed of five persons each having five different expressions resulting in a training matrix of 25 face images. These face images are DCT transformed and a block size of 16×16 is transformed using PCA to get a feature vector of 25 features.
Conclusion
A face recognition system is suggested and tested. The systems performs overall 2D-DCT on face images, chooses a block of DCT coefficients to be further transformed using PCA. The technique is simple and in-spite of the small number of used features, it can overcome variations in pose and illumination giving results that are comparable to the published techniques found in the references without any pre-processing. The recognition results illustrate that the chosen block size does not have a recognizable effect on the recognition rate.
[attachment=23184]
Introduction
Over the past few years, the user authentication is increasingly important as the security control is required everywhere. Traditionally, ID cards and passwords are popular for authentication although the security is not so reliable and convenient. Recently, biological authentication technologies (i.e Biometrics) through voice, fingerprints, iris print, retina, palm print, face..etc is playing an important role in modern personal identification systems. The face, among them, is chosen for the suggested system. The face is user friendly as people are accustomed to taking photographs. It is economic due to the low cost of cameras and computers (it does not require any special devices).
The Discrete Cosine Transform
The DCT is a popular technique in imaging and video compression, which was first applied in image compression in 1974 by Ahmed et al. In 1992, the first international standard for image compression, known as the Joint Photographic Experts Group (JPEG), was established with the DCT encoder and decoder.
The Principal Component Analysis
Eigenface is one of the most thoroughly investigated approaches to face recognition. It is also known as Karhunen-Leove expansion, eigenpicture, eigenvector, and principal component [12]. Sirovich and Kirby and Kirby et al. used PCA to efficiently represent pictures of faces. They argued that any face image can be approximately reconstructed by a small collection of weights for each face and a standard face image (eigenpicture). The weights describing each face are obtained by projecting the face image onto the eigenpicture. In mathematical terms, eigenfaces are the principal components of the distribution of faces, or the eigenvectors of the covariance matrix of the set of face images.
Recognition Results
The used data set is from the Olivetti Oracle Research Lab (The ORL database found in [14] ). The ORL database consists of 400 frontal faces; 10 tightly cropped images of 40 individuals with variations in pose, illumination, facial expressions and accessories. The size of each image is 92×112. The training set is composed of five persons each having five different expressions resulting in a training matrix of 25 face images. These face images are DCT transformed and a block size of 16×16 is transformed using PCA to get a feature vector of 25 features.
Conclusion
A face recognition system is suggested and tested. The systems performs overall 2D-DCT on face images, chooses a block of DCT coefficients to be further transformed using PCA. The technique is simple and in-spite of the small number of used features, it can overcome variations in pose and illumination giving results that are comparable to the published techniques found in the references without any pre-processing. The recognition results illustrate that the chosen block size does not have a recognizable effect on the recognition rate.