Narrow Results

In order to solve the effects of illumination changes and differences of personal features on the face recognition rate, this paper presents a new face recognition algorithm based on Gabor wavelet and Locality Preserving Projections (LPP). The problem of the Gabor filter banks with high dimensions was solved effectively, and also the shortcoming of the LPP on the light illumination changes was overcome. Firstly, the features of global image information were achieved, which used the good spatial locality and orientation selectivity of Gabor wavelet filters. Then the dimensions were reduced by utilizing the LPP, which well-preserved the local information of the image. The experimental results shown that this algorithm can effectively extract the features relating to facial expressions, attitude and other information. Besides, it can reduce influence of the illumination changes and the differences in personal features effectively, which improves the face recognition rate to 99.2%.

Machine analysis of facial emotion recognition is a challenging and an innovative research topic in human-computer intelligent interaction nowadays. The eye and the mouth regions are most essential components for facial emotion recognition. Most of the existing approaches have not utilized the eye and the mouth regions for high recognition rate. This paper proposes an approach to overcome this limitation using the eye and the mouth region-based emotion recognition using reinforced local binary patterns (LBP). The local features are extracted in each frame by using Gabor wavelet with selected scale and orientations. This feature is passed on to the ensemble classifier for detecting the location of the face region. From the signature of each pixel on the face, the eye and the mouth regions are detected using ensemble classifier. The eye and the mouth features are extracted using reinforced LBP. Multi-class Adaboost algorithm is used to select and classify these discriminative features for recognizing the emotion of the face. The developed methods are deployed on the RML, CK and FERA 2011 databases, and they exhibit significant performance improvement owing to their novel features when compared to the existing techniques.

Computer-assisted colon cancer detection on the histopathological images has become a tedious task due to its shape characteristics and other biological properties. The images acquired through the histopathological microscope may vary in magnifications for better visibility. This may change the morphological properties and hence an automated magnification independent colon cancer detection system is essential. The manual diagnosis of colon biopsy images is subjective, sluggish, laborious leading to nonconformity between histopathologists due to visual evaluation at various microscopic magnifications. Automatic detection of colon across image magnifications is challenging due to many aspects like tailored segmentation and varying features. This demands techniques that take advantage of the textural, color, and geometric properties of colon tissue. This work exhibits a segmentation approach based on the morphological features derived from the segmented region. Gabor Wavelet, Harris Corner, and DWT-LBP coefficients are extracted as it should not be dependent on the spatial domain with respect to the magnification. These features are fed to the Genetically Optimized Neural Network classifier to classify them as normal and malignant ones. Here, the genetic algorithm is used to learn the best hyper-parameters for a neural network.

The chronic and uncontrolled diabetes mellitus (DM) damages the retinal blood vessels leading to diabetic retinopathy (DR). The advanced stage of DR leads to loss of vision and subsequently blindness. The morphological changes during the progression of DR can be diagnosed using digital fundus images. The pathological changes in the retina influence the variations in pixel patterns which can be quantified using texture measures. In this paper, we have explored different texture measures namely statistical moments, gray level co-occurrence matrix (GLCM), gray level run length matrix (GLRLM), local binary pattern (LBP), laws mask energy (LME), fractal dimension (FD), fourier spectrum (FS) and Gabor wavelet to characterize and classify the normal and DR classes. We have tabulated 109 texture parameters for the normal and DR classes. Further, these features were subjected to empirical receiver operating characteristic (ROC) based ranking to select optimal feature set. The ranked nested features were fed to the support vector machine (SVM) classifier with different kernel functions to evaluate the highest performance measure using the least number of features to discriminate normal and DR classes. Our proposed system was evaluated using two different databases Kasturba Medical College Hospital (KMCH) and Tan Tock Seng Hospital (TTSH), each with 340 images (170 normal and 170 DR). We have also formulated an integrated index called as diabetic retinopathy risk index (DRRI) using selected texture features to discriminate normal and DR classes using single number. The proposed frame work can be used to help the clinicians and also for mass DR screening programs.

In this paper, a novel face descriptor, the Gabor-scale binary pattern (GSBP), is proposed to explore the neighboring relationship in spatial, frequency and orientation domains for the purpose of face recognition. In order to extract the GSBP feature, the Gabor-scale volume and the Gabor-scale vector are introduced by using a group of Gabor wavelet coefficients with a special orientation. Moreover, the Gabor-scale length pattern and the Gabor-scale ratio pattern are proposed. Compared with the existed methods, GSBP utilizes the deep relations between neighboring Gabor subimages instead of directly combining Gabor wavelet transform and local binary pattern. For estimating the performance of GSBP, we compare the proposed method with the related methods on several popular face databases, including LFW, FERET, AR, Yale and Extended YaleB databases. The experimental results show that the proposed method outperforms several popular face recognition methods.

This paper presents a pork quality evaluation method based on the hyperspectral image datasets of 96 pork samples in the range of 400–1000nm. First, through the K-medoids clustering algorithm based on manifold distance, 30 important wavelengths are selected from 753 wavelengths, and final 8 optimum wavelengths are obtained based on the discriminant value and the spectral resolution. Then, the two-dimensional Gabor wavelet transform is used to extract the eight texture features of the image under the final eight wavelengths respectively, to form a 64-dimensional features of pork quality. Finally, using the fussy C-means (FCM) algorithm based on Isomap dimension reduction, the pork quality evaluation model is constructed. The result of wavelength extraction experiments show that although there is a strong linear correlation between adjacent bands in the hyperspectral image, there is an obvious nonlinear manifold relation in the whole band. Therefore, the K-medoids clustering algorithm based on manifold distance in this paper is more reasonable than the traditional principal component analysis (PCA) in characteristic wavelength selection. According to the experiment of pork quality evaluation, two-dimensional Gabor wavelet transform can extract the texture characteristics of pork better. Compared with the FCM algorithm based on PCA, the FCM algorithm based on Isomap can better solve the high-dimensional clustering problem, and can distinguish fresh chilled meat, frozen-thawed meat and spoiled meat accurately. The study shows that hyperspectral image technology can be used in pork quality evaluation.

A study on ultrasound kidney images using proposed dominant Gabor wavelet is made for the automated diagnosis and classification of few important kidney categories namely normal, medical renal diseases and cortical cyst. The acquired images are initially preprocessed to retain the pixels of kidney region. Out of 30 Gabor wavelets, a unique dominant Gabor wavelet is determined by estimating the similarity metrics between original and reconstructed Gabor image. The Gabor features are then evaluated for each image. These derived features are mapped onto 2D feature space using k-mean clustering algorithm to group the data of similar class. The decision boundaries are formulated using linear discriminant function between the data sets of three kidney categories. A k-NN classifier module is used to identify the query input US kidney image category. The results show that the proposed dominant Gabor wavelet provides the classification efficiency of 87.33% for NR, 76.66% for MRD and 83.33% for CC. The overall classification efficiency improves by 18.89% compared to the classifier trained with features obtained by considering all the Gabor wavelets. The outputs of the proposed decision support systems are validated with medical expert to measure the actual efficiency. Also the overall discriminating ability of the systems is accessed with performance evaluation measure – f-score. It has been observed that the dominant Gabor wavelet improves the classification efficiency appreciably. Hence, the proposed method enhances the objective classification and explores the possibility of implementing a computer-aided diagnosis system exclusively for ultrasound kidney images.