Length Of Feature Vector Research Articles

Wavelet Packet Texture Descriptors Based Four-class BIRADS Breast Tissue Density Classification

Abstract The present work proposes a computer aided diagnostic (CAD) system for four-class BIRADS breast tissue density classification. The study has been carried out on a total of 480 mammograms taken from the DDSM database. The ROI has been extracted manually from each image. Each ROI has been decomposed up to 2nd level of decomposition using wavelet packet transform, resulting in 16 sub-band images for each ROI image. From each sub-band image three multi-resolution texture descriptors (i.e., mean, standard deviation and energy descriptors) are computed, resulting in texture feature vector of length 48 (16 X 3) for each ROI. The results obtained from the present work, indicate that the standard deviation and energy descriptors computed from sub-band images obtained by using wavelet packet transform with haar wavelet filter yield the highest overall classification accuracy of 73.7% for four-class breast tissue density classification. Thus it can be concluded that standard deviation and energy multi-resolution texture descriptors computed from sub-band images yielded by wavelet packet transform using haar wavelet filter contain significant information for differential diagnosis between four BIRADS breast tissue density classes.

Comparing k-means clusters on parallel Persian-English corpus

This paper compares clusters of aligned Persian and English texts obtained from k-means method. Text clustering has many applications in various fields of natural language processing. So far, much English documents clustering research has been accomplished. Now this question arises, are the results of them extendable to other languages? Since the goal of document clustering is grouping of documents based on their content, it is expected that the answer to this question is yes. On the other hand, many differences between various languages can cause the answer to this question to be no. This research has focused on kmeans that is one of the basic and popular document clustering methods. We want to know whether the clusters of aligned Persian and English texts obtained by the k-means are similar. To find an answer to this question, Mizan English-Persian Parallel Corpus was considered as benchmark. After features extraction using text mining techniques and applying the PCA dimension reduction method, the k-means clustering was performed. The morphological difference between English and Persian languages caused the larger feature vector length for Persian. So almost in all experiments, the English results were slightly richer than those in Persian. Aside from these differences, the overall behavior of Persian and English clusters was similar. These similar behaviors showed that results of k-means research on English can be expanded to Persian. Finally, there is hope that despite many differences between various languages, clustering methods may be extendable to other languages.

Color Histogram and DBC Co-Occurrence Matrix for Content Based Image Retrieval

This paper presents the integration of color histogram and DBC co-occurrence matrix for content based image retrieval. The exit DBC collect the directional edges which are calculated by applying the first-order derivatives in 0o , 45o , 90o and 135o directions. The feature vector length of DBC for a particular direction is 512 which are more for image retrieval. To avoid this problem, we collect the directional edges by excluding the center pixel and further applied the rotation invariant property. Further, we calculated the co- occurrence matrix to form the feature vector. Finally, the HSV color histogram and the DBC co-occurrence matrix are integrated to form the feature database. The retrieval results of the proposed method have been tested by conducting three experiments on Brodatz, MIT VisTex texture databases and Corel-1000 natural database. The results after being investigated show a significant improvement in terms of their evaluation measures as compared to LBP, DBC and other transform domain features.

Normalizing variations in feature vector structure in keystroke dynamics authentication systems

Usernames and passwords stubbornly remain the most prevalent authentication mechanism. Password secrecy ensures that only genuine users are granted access. If the secret is breached, impostors gain the access too. One method of strengthening password authentication is through keystroke dynamics. Keystroke dynamics algorithms typically constrain the authentication entry to one valid sequence of key presses. In this paper, we introduce the concept of event sequences. We explore the nature of variations between multiple valid key-entry sequences and propose a scheme that effectively represents these variations. We test the efficacy of the new authentication method in distinguishing users. The experimental results show that typing proficiency of individuals is not the only determining authentication factor. We show that typing sequence variations contain sufficient discriminatory information to warrant their inclusion into user authentication methods. Based on these results, we present a novel strategy to create feature vectors for keystroke dynamics-based authentication. The proposed approach ensures that the feature vector's length and structure are related only to the length of the password, independent of its content or the order of keys pressed. This normalization of feature vector structure has multiple advantages including leveraging the discriminatory power of event sequences, faster search-and-retrieval in n-graph-based authentication systems, and simplicity. The proposed authentication scheme is applicable to both static and continual authentication systems.

Remote Sensing Image Retrieval With Global Morphological Texture Descriptors

In this paper, we present the results of applying global morphological texture descriptors to the problem of content-based remote sensing image retrieval. Specifically, we explore the potential of recently developed multiscale texture descriptors, namely, the circular covariance histogram and the rotation-invariant point triplets. Moreover, we introduce a couple of new descriptors, exploiting the Fourier power spectrum of the quasi-flat-zone-based scale space of their input. The descriptors are evaluated with the UC Merced Land Use-Land Cover data set, which has been only recently made public. The proposed approach is shown to outperform the best known retrieval scores, despite its shorter feature vector length, thus asserting the practical interest of global content descriptors as well as of mathematical morphology in this context.

An efficient minutiae based geometric hashing for fingerprint database

This paper proposes an efficient indexing technique for fingerprint database using minutiae based geometric hashing. This technique consists of two stages, known as indexing and searching. For an accurate match at the time of searching, it has proposed a fixed length feature vector built from each minutia, known as Minutia Binary Pattern. Unlike any existing geometric based indexing technique, the proposed technique inserts each minutia along with the feature vector exactly once into a hash table. As a result, it reduces both computational and memory costs. Since minutiae of all fingerprint images in the database are found to be well distributed into the hash table, no rehashing is required. Experiments over FVC 2004 datasets prove the superiority of the proposed indexing technique against well known geometric based indexing techniques.

Comparative Performance Analysis of Feature(S)-Classifier Combination for Devanagari Optical Character Recognition System

this paper presents a comparative performance analysis of feature(s)-classifier combination for Devanagari optical character recognition system. For performance evaluation, three classifiers namely support vector machines, artificial neural networks and k-nearest neighbors, and seven feature extraction approaches viz. profile direction codes, transition, zoning, directional distance distribution, Gabor filter, discrete cosine transform and gradient features have been used. The first four features have been used jointly as statistical features. The performance has also been evaluated by using the combination of these feature extraction approaches. In addition, performance evaluation has also been done by varying the feature vector length of Gabor and DCT features. For training the classifiers, 7000 samples of first 70 classes (out of 942 classes), recognized in the earlier work have been used. Such a large number of classes are due to the horizontal and vertical fusion/overlapping characters. We have chosen first 70 classes as their percentage contribution out of 942 classes has found to be 96.69%. For testing, 1400 samples have been collected separately. A corpus of 25 books has been used for sample collection. Classifiers trained on different features, have been compared for performance evaluation. It has been found that support vector machines trained with Gradient features provide the classification correctness of 99.429%, and there is no significant increase in the performance with the increase in the feature vector length.

A hybrid expert system for automatic detection of voice disorders

Pathological voice analysis is a challenging task and an important area of research in voice disorder identification. Until now, the long-time acoustic (LTA) parameters are used primitively to classify the disordered voices into pathological and normal. Selection of such optimal LTA features is a disputing task. Previous researchers have used various data projection methods like principle component analysis (PCA), linear discriminant analysis (LDA) and sub-optimal searching techniques like sequential forward selection (SFS), sequential backward selection (SBS), and individual feature selection (IFS) methods for this purpose. But, these methods work efficiently for linearly separable datasets only. In order to overcome these issues, we propose a hybrid expert system in this paper, which includes the optimal selection of LTA parameters using genetic algorithm (GA), followed by non-linear classification algorithms to classify the two classes of voice samples. Nowadays, though many non-linear and high-order spectral parameters of voices have been used in this application, LTA features are scoring more importance because their clinical diagnosis is of more ease. Within this context, the GA-based feature vector quantisation combined with SVM classification is demonstrated to be more reliable, yielding 96.86% of classification accuracy for a feature vector of length 10.

Human recognition system for outdoor videos using Hidden Markov model

Human recognition is an essential requirement for human-centric surveillance, activity recognition, gait recognition etc. Inaccurate recognition of humans in such applications may leads to false alarm and unnecessary computation. In the proposed work a robust background modeling algorithm using fuzzy logic is used to detect foreground objects. Three distinct features are extracted from the contours of detected objects. An unique aggregated feature vector is formed using a fuzzy inference system by aggregating three feature vectors. To minimize computation in recognition using Hidden Markov model (HMM), the length of final feature vector is reduced using vector quantization. The proposed method is explained using five basic phases; background modeling and foreground object detection, features extraction, aggregated feature vector calculation, vector quantization, and recognition using Hidden Markov model.

Grey Scale Texture Classification Method Comparison Considering Object and Lighting Rotation

This study addresses the statistical texture features as methods for texture classification. It compares its performance on two benchmark data sets: Brodatz and PID, which permit to gain better understanding how those methods deal with texture rotation and lighting changes. Moreover, few simple feature techniques are introduced in order to compare their performance with those already known (e.g. first order features, co-occurrence matrix, run length matrix, grey-tone difference matrix, local binary pattern). Finally, exploiting structures designed in methods like co-occurrence matrices as a feature vector is suggested. The gathered results show the correct classification ratio in range of 92-100%. However, worse performance is noticeable on data set with changing lighting conditions. Moreover, the experiments prove that the introduced simple techniques classify with similar accuracy as well known methods. It is also interesting that exploiting the structures as feature vector proved to improve the classification results. Additionally, due to lower classification calculation complexity the feature vectors length have been diminished with the application of principal component analysis. This experiments showed that exploiting 95% of original information considerably reduces the feature vector length and do not influences the correct classification ratio of all tested methods.

SVM-Based Characterization of Liver Ultrasound Images Using Wavelet Packet Texture Descriptors

A system to characterize normal liver, cirrhotic liver and hepatocellular carcinoma (HCC) evolved on cirrhotic liver is proposed in this paper. The study is performed with 56 real ultrasound images (15 normal, 16 cirrhotic and 25 HCC liver images) taken from 56 subjects. A total of 180 nonoverlapping regions of interest (ROIs), i.e. 60 from each image class, are extracted by an experienced participating radiologist. The multiresolution wavelet packet texture descriptors, i.e. mean, standard deviation and energy features, are computed from all 180 ROIs by using various compact support wavelet filters including Haar, Daubechies (db4 and db6), biorthogonal (bior3.1,bior3.3 and bior4.4), symlets (sym3 and sym5) and coiflets (coif1 and coif2). It is observed that a combined texture descriptor feature vector of length 48 consisting of 16 mean, 16 standard deviation and 16 energy features estimated from all 16 subband feature images (wavelet packets) obtained by second-level decomposition with two-dimensional wavelet packet transform by using Haar wavelet filter gives the best characterization performance of 86.6%. Feature selection by genetic algorithm-support vector machine method increased the classification accuracy to 88.8% with sensitivity of 90% for detecting normal and cirrhotic cases and sensitivity of 86.6% for HCC cases. Considering limited sensitivity of B-mode ultrasound for detecting HCCs evolved on cirrhotic liver, the sensitivity of 86.6% for HCC lesions obtained by the proposed computer-aided diagnostic system is quite promising and suggests that the proposed system can be used in a clinical environment to support radiologists in lesion interpretation.

Facial expression recognition using tracked facial actions: Classifier performance analysis

In this paper, we address the analysis and recognition of facial expressions in continuous videos. More precisely, we study classifiers performance that exploit head pose independent temporal facial action parameters. These are provided by an appearance-based 3D face tracker that simultaneously provides the 3D head pose and facial actions. The use of such tracker makes the recognition pose- and texture-independent. Two different schemes are studied. The first scheme adopts a dynamic time warping technique for recognizing expressions where training data are given by temporal signatures associated with different universal facial expressions. The second scheme models temporal signatures associated with facial actions with fixed length feature vectors (observations), and uses some machine learning algorithms in order to recognize the displayed expression. Experiments quantified the performance of different schemes. These were carried out on CMU video sequences and home-made video sequences. The results show that the use of dimension reduction techniques on the extracted time series can improve the classification performance. Moreover, these experiments show that the best recognition rate can be above 90%.

Contourlet-S Retrieval Algorithm Using Absolute Mean Energy and Kurtosis Features

Contourlet transform has better performance in directional information representation than wavelet transform and has been studied by many researchers in retrieval systems and has been shown that it is superior to wavelet ones at retrieval rate. In order to improve the retrieval rate further, a contourlet-S transform based texture image retrieval system was proposed in this paper. In the system, the contourlet transform was constructed by non-subsampled Laplacian pyramid cascaded by critical subsampled directional filter banks, sub-bands absolute mean energy and kurtosis in contourlet-S domain are cascaded to form feature vectors, and the similarity metric is Canberra distance. Experimental results on 109 brodatz texture images show that using the features cascaded by absolute mean and kurtosis can lead to a higher retrieval rate than the combination of standard deviation and absolute mean which is most commonly used today under same dimension of feature vectors. contourlet-S transform based image retrieval system is superior to those of the original contourlet transform, non-subsampled contourlet system under the same system structure with same length of feature vectors, retrieval time and memory needed, contourlet-S decomposition structure parameters can make significant effects on retrieval rates, especially scale number.

Semi-Subsampled Contourlet Retrieval Algorithm Using Three Statistical Features

In order to improve the retrieval rate of contourlet transform retrieval system, a semi-subsampled contourlet transform based texture image retrieval system was proposed. In the system, the contourlet transform was constructed by non-subsampled Laplacian pyramid cascaded by critical subsampled directional filter banks, sub-bands standard deviation, absolute mean energy and kurtosis in semi-subsampled contourlet domain are cascaded to form feature vectors, and the similarity metric is Canberra distance. Experimental results on 109 brodatz texture images show that using the three cascaded features can lead to a higher retrieval rate than the combination of standard deviation and absolute mean which is most commonly used today under same dimension of feature vectors. Semi-subsampled contourlet transform based image retrieval system is superior to those of the original contourlet transform, non-subsampled contourlet system under the same system structure with same length of feature vectors, retrieval time and memory needed, decomposition structure parameters can also make significant effects on retrieval rates, especially scale number.

Construction of Fisheye Lens Inverse Perspective Mapping Model and Its Applications of Obstacle Detection

In this paper, we develop a vision based obstacle detection system by utilizing our proposed fisheye lens inverse perspective mapping (FLIPM) method. The new mapping equations are derived to transform the images captured by the fisheye lens camera into the undistorted remapped ones under practical circumstances. In the obstacle detection, we make use of the features of vertical edges on objects from remapped images to indicate the relative positions of obstacles. The static information of remapped images in the current frame is referred to determining the features of source images in the searching stage from either the profile or temporal IPM difference image. The profile image can be acquired by several processes such as sharpening, edge detection, morphological operation, and modified thinning algorithms on the remapped image. The temporal IPM difference image can be obtained by a spatial shift on the remapped image in the previous frame. Moreover, the polar histogram and its post-processing procedures will be used to indicate the position and length of feature vectors and to remove noises as well. Our obstacle detection can give drivers the warning signals within a limited distance from nearby vehicles while the detected obstacles are even with the quasi-vertical edges.

Svm PRAT: SVM-based Protein Residue Annotation Toolkit

BackgroundOver the last decade several prediction methods have been developed for determining the structural and functional properties of individual protein residues using sequence and sequence-derived information. Most of these methods are based on support vector machines as they provide accurate and generalizable prediction models.ResultsWe present a general purpose protein residue annotation toolkit (svmPRAT) to allow biologists to formulate residue-wise prediction problems. svmPRAT formulates the annotation problem as a classification or regression problem using support vector machines. One of the key features of svmPRAT is its ease of use in incorporating any user-provided information in the form of feature matrices. For every residue svmPRAT captures local information around the reside to create fixed length feature vectors. svmPRAT implements accurate and fast kernel functions, and also introduces a flexible window-based encoding scheme that accurately captures signals and pattern for training effective predictive models.ConclusionsIn this work we evaluate svmPRAT on several classification and regression problems including disorder prediction, residue-wise contact order estimation, DNA-binding site prediction, and local structure alphabet prediction. svmPRAT has also been used for the development of state-of-the-art transmembrane helix prediction method called TOPTMH, and secondary structure prediction method called YASSPP. This toolkit developed provides practitioners an efficient and easy-to-use tool for a wide variety of annotation problems.Availability: http://www.cs.gmu.edu/~mlbio/svmprat

Optimal feature selection for the assessment of vocal fold disorders

Unilateral vocal fold paralysis, vocal fold polyp, and vocal fold nodules are the most common types of neurogenic and organic vocal disorders. This article aims to distinguish these types of vocal diseases into four different classes for the purpose of automatic screening. Firstly, the reconstructed signal at each wavelet packet decomposition sub-band in five levels of decomposition with mother wavelet of (db10) is used to extract the nonlinear features of self-similarity and approximate entropy. Also, wavelet packet coefficients are used to measure energy and Shannon entropy features at different spectral sub-bands. Consequently, to find a discriminant feature vector, three different methods have been applied: Davies–Bouldin (DB) criteria, genetic algorithm (GA) with the fitness functions of support vector machine's (SVM) and k-nearest neighbor's (KNN) recognition rates. Finally, obtained feature vectors have been passed on to SVM and KNN classifiers. The results show that a feature vector of length 12 obtained by the optimization method of GA with the fitness function of SVM's recognition rate fed to SVM classifier achieves the highest classification accuracy of 91%. Furthermore, nonlinear features play an important role in pathological voice classification by participating rate of approximately 67% in the optimal feature vector.

Descriptors for image-based fingerprint matchers

This paper focuses on the use of image-based techniques in fingerprint verification. A detailed review of the existing literature is provided by classifying existing methods on the basis of their alignment procedure and discussing the most salient approaches and their pros and cons. Even if, at present, the image-based techniques do not gain performance comparable with that obtained by the best minutiae-based approaches, several good reasons can be listed to support the research on image-based approaches: the possibility of using additional features in combination with minutiae to improve verification performance, the availability of a fixed length feature vector which makes these approaches suitable to be indexed, to be coupled with a learning system or to be combined with tokenised random number in a two factor authentication system (Biohashing). In this work we compare several texture-based descriptors for fingerprints and propose a novel image-based fingerprint matcher based on the minutiae alignment. In this approach, the feature extraction is performed locally on a decomposition of the fingerprint in several overlapping sub-windows considering the following measures: Gabor filters descriptors, invariant local binary patterns and histogram of gradients. Moreover, we propose to perform a supervised selection of a small subset of descriptors, in order to reduce the dimensionality of the feature set and discarding the less discriminative features. Extensive experiments conducted over the four FVC2002 fingerprint databases using a blind testing protocol show that the proposed system dramatically outperforms the other image-based fingerprint matchers proposed in the literature. Moreover, a further experiment conducted on a set of images reconstructed from ISO templates show that, differently to the minutiae-based approaches, our image-based matcher cannot be faked with the sole knowledge of the minutiae position and orientation, at least the original orientation image is required in order have a chance of performing a successful attack.

Archival image indexing with connectivity features using randomized masks

Connectivity properties capture a natural spatial feature of a binary image. Albeit they are easy to compute, more often than not, they alone fail to provide a good characterization of the image because of the fact that several different images may have the same connectivity features. Typically, other types of parameters, e.g., moments, are used to augment the connectivity feature for efficient indexing and retrieval purposes. In this work, an alternative approach is proposed. Instead of considering other diverse features, which are computationally intensive, only the connectivity features of the image are used iteratively using a novel concept of spatial masking. For this purpose, a greedy algorithm for constructing a spatial feature vector of variable length for a binary image, is proposed. The algorithm is based on XOR-ing the image bit-plane with a few pseudo-random synthetic masks, and its novelty lies in computing the feature vector iteratively, depending on the size and diversity of the image database. The classical Euler number and the two primary connectivity features from which it is derived, namely, the number of connected components and the number of holes, are used to finally generate a unique feature vector for each binary image in the database using a fuzzy membership function customized for the given database. The method is particularly suitable for large-sized image archives of a digital library, where each image contains one or more objects. It is found to converge within only three iterations for a postal stamp database consisting of 2598 images, and also for a logo database of 1034 images. A data structure called discrimination tree has been introduced for supporting efficient storage and indexing of the images using the above feature vector.

A procedure for determining time-of-day break points for coordinated actuated traffic signal systems

This paper presents the development and evaluation of a procedure for determining optimal break points for time-of-day based coordinated actuated traffic signal operations. The proposed procedure uses a feature vector of optimal cycle length per time interval instead of traffic volume itself. Initial break points determined by the proposed feature vector are used in the greedy search algorithm to obtain optimal break points. By using the greedy search algorithm the number of evaluations in the search are dramatically reduced when compared to an exhaustive search or other common heuristic search methods such as a genetic algorithm. The proposed procedure was evaluated using a hypothetical network consisting of four signalized intersections and the results indicated that the proposed procedure effectively improved the performance of the coordinated actuated signal control. In addition, sensitivity analyses results on randomly varying demand conditions up to ±20% and randomly increased demand conditions up to 30% indicated that the newly developed break points were robust for such varying demand fluctuations.