Kinds Of Feature Vectors Research Articles

Tree Species Classification Using Optimized Features Derived from Light Detection and Ranging Point Clouds Based on Fractal Geometry and Quantitative Structure Model

Tree species classification is a ubiquitous task in the forest inventory field. Only directly measured feature vectors have been applied to most existing methods that use LiDAR technology for tree species classification. As a result, it is difficult to obtain a satisfactory tree species classification performance. To solve this challenge, the authors of this paper developed two new kinds of feature vectors, including fractal geometry-based feature vectors and quantitative structural model (QSM)-based feature vectors. In terms of fractal geometry, both two fractal parameters were extracted as feature vectors for reflecting how tree architecture is distributed in three-dimensional space. In terms of QSM, the ratio of length change and the ratio of radius change of different branches were extracted as feature vectors. To reduce the feature vector dimensionality and explore valuable feature vectors, feature vector dimension reduction was conducted using the classification and regression tree (CART). Five hundred and sixty-eight individual trees with five tree species were selected for evaluating the performance of the developed feature vectors. The experimental results indicate that the tree species of Fagus sylvatica achieved the highest overall accuracy, which is 98.06%, while Quercus petraea obtained the lowest overall accuracy, which is 96.65%. Four other classical supervised learning methods were adopted for comparison. The comparison result indicates that the proposed method outperformed the other four supervised learning methods no matter which accuracy indicator was adopted. In comparison with the relevant method, the eight feature vectors developed in this paper also performed much better. This indicates that the fractal geometry-based feature vectors and QSM-based feature vectors developed in this paper can effectively improve the performance of tree species classification.

Multi-modal facial expression feature based on deep-neural networks

Emotion recognition based on facial expression is a challenging research topic and has attracted a great deal of attention in the past few years. This paper presents a novel method, utilizing multi-modal strategy to extract emotion features from facial expression images. The basic idea is to combine the low-level empirical feature and the high-level self-learning feature into a multi-modal feature. The 2-dimensional coordinate of facial key points are extracted as low-level empirical feature and the high-level self-learning feature are extracted by the Convolutional Neural Networks (CNNs). To reduce the number of free parameters of CNNs, small filters are utilized for all convolutional layers. Owing to multiple small filters are equivalent of a large filter, which can reduce the number of parameters to learn effectively. And label-preserving transformation is used to enlarge the dataset artificially, in order to address the over-fitting and data imbalance of deep neural networks. Then, two kinds of modal features are fused linearly to form the facial expression feature. Extensive experiments are evaluated on the extended Cohn–Kanade (CK+) Dataset. For comparison, three kinds of feature vectors are adopted: low-level facial key point feature vector, high-level self-learning feature vector and multi-modal feature vector. The experiment results show that the multi-modal strategy can achieve encouraging recognition results compared to the single modal strategy.

Estimating a person's age from walking over a sensor floor

Ageing has an effect on many parameters of the physical condition, and one of them is the way a person walks. This property, the gait pattern, can unintrusively be observed by letting people walk over a sensor floor. The electric capacitance sensors built into the floor deliver information about when and where feet get into close proximity and contact with the floor during the phases of human locomotion. We processed gait patterns recorded this way by extracting a feature vector containing the discretised distribution of occurring geometrical extents of significant sensor readings. This kind of feature vector is an implicit measure encoding the ratio of swing-to stance phase timings in the gait cycle and representing how cleanly the leg swing is performed. We then used the dataset to train a Multi-Layer Perceptron to perform regression with the age of the person as the target value, and the feature vector as input. With this method and a dataset size of 142 persons recorded, we achieved a mean absolute error of approximately 10 years between the true age and the estimated age of the person. Considering the novelty of our approach, this is an acceptable result. The combination of a floor sensor and machine learning methods for interpreting the sensor data seems promising for further research and applications in care and medicine.

Feature Sensitive Label Fusion With Random Walker for Atlas-Based Image Segmentation.

In this paper, a novel label fusion method is proposed for brain magnetic resonance image segmentation. This label fusion method is formulated on a graph, which embraces both label priors from atlases and anatomical priors from target image. To represent a pixel in a comprehensive way, three kinds of feature vectors are generated, including intensity, gradient, and structural signature. To select candidate atlas nodes for fusion, rather than exact searching, randomized k-d tree with spatial constraint is introduced as an efficient approximation for high-dimensional feature matching. Feature sensitive label prior (FSLP), which takes both the consistency and variety of different features into consideration, is proposed to gather atlas priors. As FSLP is a non-convex problem, one heuristic approach is further designed to solve it efficiently. Moreover, based on the anatomical knowledge, parts of the target pixels are also employed as the graph seeds to assist the label fusion process, and an iterative strategy is utilized to gradually update the label map. The comprehensive experiments carried out on two publicly available databases give results to demonstrate that the proposed method can obtain better segmentation quality.

Performance Evaluation on Feature Modeling of Halftone Image Texture

Three statistics methods, gray-level co-occurrence matrix, autocorrelation function and spectrum statistics, were used to extract feature vector of various halftone images for halftone image classification. The classification perfor- mances of three kinds of feature vectors were assessed by three classifiers: radial basis function neural network, least mean square and principal component analysis. Experimental results showed the autocorrelation function is better than other two methods for classification of halftone image. It indicated the best classification performance when the parameter K=64 and L=8.

Comparison of organs’ shapes with geometric and Zernike 3D moments

The morphological similarity of organs is studied with feature vectors based on geometric and Zernike 3D moments. It is particularly investigated if outliers and average models can be identified. For this purpose, the relative proximity to the mean feature vector is defined, principal coordinate and clustering analyses are also performed. To study the consistency and usefulness of this approach, 17 livers and 76 hearts voxel models from several sources are considered.In the liver case, models with similar morphological feature are identified. For the limited amount of studied cases, the liver of the ICRP male voxel model is identified as a better surrogate than the female one. For hearts, the clustering analysis shows that three heart shapes represent about 80% of the morphological variations. The relative proximity and clustering analysis rather consistently identify outliers and average models.For the two cases, identification of outliers and surrogate of average models is rather robust. However, deeper classification of morphological feature is subject to caution and can only be performed after cross analysis of at least two kinds of feature vectors. Finally, the Zernike moments contain all the information needed to re-construct the studied objects and thus appear as a promising tool to derive statistical organ shapes.

A Novel Multimodal Biometric System based on Iris and Face

With the wide application, the recognition accuracy of unimodal biometric systems has to contend with a variety of problems such as background noise, signal noise and distortion, and environment or device variations. This paper combines face and iris features for developing a multimode biometric system, which is able to diminish the above mentioned problems of unimodal biometrics as well as to improve the performance of authentication system. We adopt an efficient feature-level fusion scheme for iris feature and face feature in series, and normalize the original features of iris and face using zscore model to eliminate the unbalance of magnitude and the distribution between two different kinds of feature vectors. Fisher discriminant analysis is used to select the distinguishing character of the fused feature in order to further enhance the performance of the system. The proposed algorithm is tested using CASIA iris database and two face databases (ORL database and Yale database). Experimental results reveal the multimodal biometrics verification is much more reliable and precise than unimodal biometric system.

Diagnostics of Filtered Analog Circuits with Tolerance Based on LS-SVM Using Frequency Features

Most researchers have used the optimal wavelet coefficients or wavelet energy indicators from the time-domain response of analog circuits to train support vector machines (SVMs) to diagnose faults. In this study, we have proposed two kinds of feature vectors from frequency response data of a filter system to train least squares SVM (LS-SVM) to diagnose faults. The first is defined as the conventional frequency feature vector, which includes the center frequency and the maximum frequency response. The second is a new wavelet feature vector that is composed of the mean and standard deviation of wavelet coefficients. Different feature vectors' combination and normalization are also discussed in the paper. The results from the simulation data and the real data for two filters showed the following: (1) The proposed method has better diagnostic accuracy than the traditional methods that were based only on the optimal wavelet coefficients or wavelet energy indicators. (2) The diagnostic accuracies using the combined feature vectors were better than those using only the conventional frequency feature vectors or wavelet feature vectors. (3) The best accuracy from using the conventional frequency feature vectors was better than that from using wavelet feature vectors. The proposed method can be extended to diagnostics of other analog circuits that are determined by their frequency characteristics.

Multimodal Biometric System Using Face-Iris Fusion Feature

With the wide application, the performance of unimodal biometrics systems has to contend with a variety of problems such as background noise, signal noise and distortion, and environment or device variations. Therefore, multimodal biometric systems are proposed to solve the above mentioned problems. This paper proposed a novel multimodal biometric system using face-iris fusion feature. Face feature and iris feature are first extracted respectively and fused in feature-level. However, existing feature level schemes such as sum rule and weighted sum rule are inefficient in complicated condition. In this paper, we adopt an efficient feature-level fusion scheme for iris and face in series. The algorithm normalizes the original features of iris and face using z-score model to eliminate the unbalance in the order of magnitude and the distribution between two different kinds of feature vectors, and then connect the normalized feature vectors in serial rule. The proposed algorithm is tested using CASIA iris database and two face databases (ORL database and Yale database). Experimental results show the effectiveness of the proposed algorithm.

Prediction of protein–protein interaction types using the decision templates based on multiple classier fusion

Protein–protein interactions (PPIs) play a key role in many cellular processes, such as the regulation of enzymes, signal transduction or mediating the adhesion of cells. Knowing about the multitude of PPIs that allow the cell to function can help the biological scientist understand the molecular machinery of the cell. Unfortunately, it is both time-consuming and expensive to do so solely based on experiments due to the nature of the problem whose complexity is obviously overwhelming, just like the fact that “life is complicated”. Therefore, developing computational approaches for predicting PPIs, binding sites and interaction types would be of significant value in this regard. In this paper, we propose a novel method for predicting the types of PPIs based on decision templates. First, we introduce the concept of the tensor product to construct three kinds of feature vectors which are the amino acid composition tensor product, the residue multi-scale conservation energy tensor product and the secondary structure content tensor product, then the correlation-based feature selection method was also introduced to reduce the dimensionality of these feature vectors. So, the protein pair can be represented by our three new kinds of feature vectors and Zhu’s six kinds of feature vectors. The nine kinds of feature vectors are further taken as the inputs of individual support vector machine classifier respectively, and the outputs of these classifiers are aggregated with decision templates. The overall success rate obtained by jackknife cross-validation was 90.95%, indicating our method is very promising for predicting PPI types, might become a useful vehicle for studying the network biology in the post-genomic era.

An Ego-Motion Detection System Employing Directional-Edge-Based Motion Field Representations

In this paper, a motion field representation algorithm based on directional edge information has been developed. This work is aiming at building an ego-motion detection system using dedicated VLSI chips developed for real time motion field generation at low powers [1], [2]. Directional edge maps are utilized instead of original gray-scale images to represent local features of an image and to detect the local motion component in a moving image sequence. Motion detection by edge histogram matching has drastically reduced the computational cost of block matching, while achieving a robust performance of the ego-motion detection system under dynamic illumination variation. Two kinds of feature vectors, the global motion vector and the component distribution vectors, are generated from a motion field at two different scales and perspectives. They are jointly utilized in the hierarchical classification scheme employing multiple-clue matching. As a result, the problems of motion ambiguity as well as motion field distortion caused by camera shaking during video capture have been resolved. The performance of the ego-motion detection system was evaluated under various circumstances, and the effectiveness of this work has been verified.

An adaptive feature fusion framework for multi-class classification based on SVM

An adaptive feature fusion framework is proposed for multi-class classification based on SVM. In a similar manner of one-versus-all (OVA), one of the multi-class SVM schemes, the proposed approach decomposes a multi-class classification into several binary classifications. The main difference lies in that each classifier is created with the most suitable feature vectors to discriminate one class from all the other classes. The feature vectors of the unknown samples are selected by each classifier adaptively such that recognition is fulfilled accordingly. In addition, novel evaluation criterions are defined to deal with the frequent small-number sample problems. A writer recognition experiment is carried out to accomplish this framework with three kinds of feature vectors: texture, structure and morphological features. Finally, the performance of the proposed approach is illustrated as compared with the OVA by applying the same feature vectors for all classes.

Multi-level spherical moments based 3D model retrieval

In this paper a novel 3D model retrieval method that employs multi-level spherical moment analysis and relies on voxelization and spherical mapping of the 3D models is proposed. For a given polygon-soup 3D model, first a pose normalization step is done to align the model into a canonical coordinate frame so as to define the shape representation with respect to this orientation. Afterward we rasterize its exterior surface into cubical voxel grids, then a series of homocentric spheres with their center superposing the center of the voxel grids cut the voxel grids into several spherical images. Finally moments belonging to each sphere are computed and the moments of all spheres constitute the descriptor of the model. Experiments showed that Euclidean distance based on this kind of feature vector can distinguish different 3D models well and that the 3D model retrieval system based on this arithmetic yields satisfactory performance.

Improving Classification Using a Tree Structured Neural Network

This article introduces a tree structured network for improving the performance of the feedforward neural network FN classifier. The building blocks of the tree are the feedforward neural network with backpropagation learning scheme and the simple logical OR neural network ORNN. The confusion matrix CM, resulting from some preliminary experiments, is used to divide the considered patterns into groups in several primary stages until no more grouping could be obtained. The proposed structure can be used for any pattern classification problem. In this article, the testing environment is the isolated handwritten Arabic character set, which is a problem of reasonable complexity. Two simple kinds of feature vectors are used to represent characters before the FN. The use of the proposed tree structure improved the recognition accuracy from 80% for the single FN to 97.7% using the same simple set of features. The results show that with the proposed method, better classification results could be obtained without having to introduce more complex features.