A Hybrid Feature Extraction Technique for Face Recognition

Author(s):  
Aayushi Sharma ◽  
◽  
Sheetal Chhabra ◽  

The advancement of image editing software tools in the image processing field has led to an exponential increase in the manipulation of the images. Subjective differentiation of original and manipulated images has become almost impossible. This has kindled the interest among researchers to develop algorithms for detecting the forgery in the image. ImageSplicing, Copy-Move and Image Retouching are the most common image forgery techniques. The existing methods to detect image forgery has drawbacks like false detection, high execution time and low accuracy rate. Considering these issues, this work proposes an efficient method for detection of image forgery. Initially, bilateral filter is used to remove the noise in pre-processing, Chan-Vese Segmentation algorithm is used to detect the clumps from the filtered image utilizing both intensity and edgeinformation, followed by hybrid feature extraction technique. Hybrid feature extraction technique comprises of Dual Tree Complex-Discrete Wavelet Transform (DWT), Principal Component Analysis (PCA) and Gray-Level-Co-Occurrence Matrix (GLCM). The DWT has dual-tree complex wavelet transform with important properties, it is nearly shift invariant and directionally selective in two and higher dimensions. Principal Component Analysis (PCA) finds the eigenvectors of a covariance matrix with the highest eigenvalues and uses these values to project the data into a new subspace of equal or less dimensions. Gray-Level-Co-Occurrence Matrix (GLCM) extracts the Feature values such as energy, entropy, homogeneity, standard deviation, variance, contrast, correlation and mean. Classification is done based on the texture values of training dataset and testing dataset using Multi Class-Support Vector Machine (SVM). The performance analysis is done based on the True positive, False positive and True negative values. The experimental results obtained using the proposed technique shows a better performance compared to the existing KNN classifier model.


Author(s):  
K. RUBA SOUNDAR ◽  
K. MURUGESAN

Face recognition plays a vital role in authentication, monitoring, indexing, access control and other surveillance applications. Much research on face recognition with various feature based approaches using global or local features employing a number of similarity measurement techniques have been done earlier. Feature based approaches using global features can effectively preserve only the Euclidean structure of face space, that suffer from lack of local features which may play a major role in some applications. On the other hand, wtih local features only the face subspace that best detects the essential face manifold structure is obtained and it also suffers loss in global features which may also be important in some other applications. Measuring similarity or distance between two feature vectors is also a key step for any pattern matching application. In this work, a new combined approach for recognizing faces that integrates the advantages of the global feature extraction technique by Linear Discriminant Analysis (LDA) and the local feature extraction technique by Locality Preserving Projections (LPP) with correlation based similarity measurement technique has been discussed. This has been validated by performing various experiments and by making a fair comparison with conventional methods.


Author(s):  
Razia Zia ◽  
Pervez Akhtar ◽  
Arshad Aziz ◽  
Dur E Shahwar Kundi

The field of medical image classification has been one of the most attention-gaining research areas in the recent times due to the increasing demand for an efficient tool that can help doctors in making quick and correct diagnoses. In this paper, a hybrid feature extraction technique is proposed, which is based on discrete wavelet transform (DWT), non-subsampled contourlet transform (NSCT) and isotropic gray level co-occurrence matrix (GLCM) for the categorization of grade II, III, and IV gliomas. The proposed method was applied on a dataset of 93 MRI brain images containing three glioma grades (23 grade II, 45 grade III, and 25 grade IV). The efficiency of proposed methodology is evaluated in terms of classification accuracy, sensitivity and specificity. The highest accuracy of [Formula: see text] for grade III, sensitivity of [Formula: see text] and specificity of [Formula: see text] were achieved in case of grade II.


Sign in / Sign up

Export Citation Format

Share Document