Geometric Feature Extraction Research Articles

Visual detection of construction and demolition waste using multi-sensor fusion

In construction and demolition waste (CDW), there are materials that are difficult to separate by density and magnetic characteristics alone. To realise a finer separation of these difficult-to-separate materials, a multi-sensor fusion method for the visual detection of CDW is proposed. This method consists of a red–green–blue–depth (RGBD) data collection method with high precision in the height direction and a multimodal data processing algorithm. To obtain comprehensive features accurately and efficiently, the proposed method uses two independent MobileNet branches to extract colour visual features and geometric features separately, and an encoder is used to control the frequency of the two models in real time to better match the two data types. The proposed method makes full use of the height data to filter out the background of the conveyor belt and the invalid areas generated by the region proposal networks to improve its robustness. Experiments show that the accuracy of the proposed visual detection method is over 90%. Additionally, computational efficiency reaches 1.67 frames per second (fps) on a central processing unit (CPU) and 30 fps on a graphics processing unit. The detection efficiency when running on the CPU could reach 25 000 CDW items per hour.

An implementation of sign language alphabet hand posture recognition using geometrical features through artificial neural network (part 2)

In the sign language alphabet, several hand signs are in use. Automatic recognition of performed hand signs can facilitate the communication between hearing and none hearing people. This framework proposes hand posture recognition of the American Sign Language alphabet based on a neural network (NN) which works on geometrical feature extraction of the hand. The user’s hand is captured by a 3D depth-based sensor camera. Consequently, the hand is segmented according to the depth features. The proposed system is called ‘Depth-based Geometrical Sign Language Recognition’ (DGSLR). The DGSLR adopted an easier hand segmentation approach, which is further used in other segmentation applications. The proposed geometrical feature extraction framework improves the accuracy of recognition due to unchangeable features against hand orientation or rotation compared to Discrete Cosine Transform (DCT) and Moment Invariant. As a support vector machine (SVM) is a type of artificial neural network (ANN), it is used to drive desired outcomes. Since there are 26 different signs in the Sign Language alphabet, a multi-class SVM versus a single SVM classifier with 26 classes by an RBF kernel was used to validate each class. The proposed framework is proficient to hand posture recognition and provides an accuracy of up to 96.78%. The findings of the iterations demonstrated that the combination of the extracted features resulted in a better accuracy rate in the recognition process in the classification step.

Muscle fatigue analysis in isometric contractions using geometric features of surface electromyography signals

In this study, an attempt has been made to differentiate the muscle nonfatigue and fatigue conditions using geometric features of surface Electromyography (sEMG) signals. For this purpose, a new framework is proposed that consists of Fourier descriptor based shape representation and geometric feature extraction. The sEMG signals are acquired from biceps brachii muscle of 25 healthy adult volunteers in isometric contractions. The signals associated with nonfatigue and fatigue conditions are preprocessed and subjected to discrete Fourier transform. The Fourier coefficients are scattered in the complex plane and the envelope is computed using α-shape method. The boundary of the resultant shape represents the Fourier descriptors. The geometric features namely centroid, moments, perimeter, area, circularity, convexity, average bending energy, major axis length, eccentricity and ellipse variance are extracted from the shape. The results show that seven out of twelve features have statistically significant (p < 0.001) difference between the two conditions. The five features namely major axis length, area, perimeter, second order moment and central moment are considered for muscle fatigue classification using k-nearest neighbor, naïve Bayes, decision tree and multilayer perceptron (MLP). Among these classifiers, maximum accuracy of 86 % is achieved with MLP based detection model. Therefore, it appears that the geometric features of sEMG signals could be useful in the detection of muscle fatigue condition in clinical diagnosis, workplace and rehabilitation.

Covariance Kalman Geometric Graph Based Feature Extraction And Bernoulli Kernel Classifier For Plant Leaf Disease Prediction

As far as the agricultural domain is concerned, one of the most hot research areas of analysis is accurate prediction of leaf disease from the leaf images of a plant. The prediction of agricultural plant diseases bymeans of the image processing techniques will hence reduce the dependence on the farmers to safeguard their agricultural land and also their products. However, with the presence of noise, the leaf disease prediction is said to be hindered. To address this issue, in this paper, Covariance Kalman Geometric Graph-basedBernoulliClassifier (CKGG-BC) for Plant leaf disease prediction is proposed. The CKGG-BC method is split into three parts. To start with the plant leaf image provided as input, the Covariance Kalman Filtered Preprocessing modelintroduced for the image enhancement. Second, Geometric Graph-based Segmented Co-occurrence Feature Extraction model is applied to the preprocessed image to accurately segment the infected leaf areas and followed by which extracting the accurate infected leaf areas. Finally, Bernoulli Online Multiple Kernel Learning Classifier is applied for accurate plant leaf disease prediction with minimum classification error. The proposed method provides a significant refinement with respect to state-of-the-art methods. Even under complex background conditions, i.e., in the presence of noise, the averageaccuracy of the proposed method is said to be improved and hence paves mechanism for prediction of plant leaf disease in a significant manner. Experimentalresults exhibit the effectiveness of the proposed method in terms of computational overhead, accuracy, true positive rate and classification error respectively.

Multiscale geometric feature extraction for high-dimensional and non-Euclidean data with applications

A method for extracting multiscale geometric features from a data cloud is proposed and analyzed. Based on geometric considerations, we map each pair of data points into a real-valued feature function defined on the unit interval. Further statistical analysis is then based on the collection of feature functions. The potential of the method is illustrated by different applications, including classification and anomaly detection. Connections to other concepts, such as random set theory, localized depth measures and nonlinear dimension reduction, are also explored.

First validation of GOME-2/MetOp absorbing aerosol height using EARLINET lidar observations

Abstract. The aim of this study is to investigate the potential of the Global Ozone Monitoring Experiment-2 (GOME-2) instruments, aboard the Meteorological Operational (MetOp)-A, MetOp-B and MetOp-C satellite programme platforms, to deliver accurate geometrical features of lofted aerosol layers. For this purpose, we use archived ground-based lidar data from stations available from the European Aerosol Research Lidar Network (EARLINET) database. The data are post-processed using the wavelet covariance transform (WCT) method in order to extract geometrical features such as the planetary boundary layer (PBL) height and the cloud boundaries. To obtain a significant number of collocated and coincident GOME-2 – EARLINET cases for the period between January 2007 and September 2019, 13 lidar stations, distributed over different European latitudes, contributed to this validation. For the 172 carefully screened collocations, the mean bias was found to be −0.18 ± 1.68 km, with a near-Gaussian distribution. On a station basis, and with a couple of exceptions where very few collocations were found, their mean biases fall in the ± 1 km range with an associated standard deviation between 0.5 and 1.5 km. Considering the differences, mainly due to the temporal collocation and the difference, between the satellite pixel size and the point view of the ground-based observations, these results can be quite promising and demonstrate that stable and extended aerosol layers as captured by the satellite sensors are verified by the ground-based data. We further present an in-depth analysis of a strong and long-lasting Saharan dust intrusion over the Iberian Peninsula. We show that, for this well-developed and spatially well-spread aerosol layer, most GOME-2 retrievals fall within 1 km of the exact temporally collocated lidar observation for the entire range of 0 to 150 km radii. This finding further testifies for the capabilities of the MetOp-borne instruments to sense the atmospheric aerosol layer heights.

Extraction Method of Microscopic Image Feature of Cotton Fiber Cross Section

The micro-section image analysis is an effective method to measure cotton fiber maturity directly. However, the accuracies of the existing image segmentation and contour extraction algorithms were limited by the extraction of cotton fiber cross-section features. Therefore, the paper first binarized the cotton fiber cross-section micro image with the image processing software programmed by VC++, and then extracted the contours of its outer layer, inner layer and cavity, and finally calculated its geometric features. Based on statistics of these geometric features, a judgment model was established to remove the low quality cross sections such as pseudo-cross sections and separate adhesion cross sections, so as to improve the geometric feature extraction effect and the maturity estimation accuracy of cotton fiber cell. The results showed that, compared with the artificial subjective judgment, the cotton fiber cross-section feature extraction method proposed in this paper could not only solve the problem of unable to determine the geometric characteristics caused by cotton fiber adhesion and other interference factors, but also reduce the identification error rate from 0.073 to 0.038, so as to better judge the quality of cotton fiber at harvest time, and infer the possible influencing factors.

Facial geometric feature extraction based emotional expression classification using machine learning algorithms.

Emotion plays a significant role in interpersonal communication and also improving social life. In recent years, facial emotion recognition is highly adopted in developing human-computer interfaces (HCI) and humanoid robots. In this work, a triangulation method for extracting a novel set of geometric features is proposed to classify six emotional expressions (sadness, anger, fear, surprise, disgust, and happiness) using computer-generated markers. The subject’s face is recognized by using Haar-like features. A mathematical model has been applied to positions of eight virtual markers in a defined location on the subject’s face in an automated way. Five triangles are formed by manipulating eight markers’ positions as an edge of each triangle. Later, these eight markers are uninterruptedly tracked by Lucas- Kanade optical flow algorithm while subjects’ articulating facial expressions. The movement of the markers during facial expression directly changes the property of each triangle. The area of the triangle (AoT), Inscribed circle circumference (ICC), and the Inscribed circle area of a triangle (ICAT) are extracted as features to classify the facial emotions. These features are used to distinguish six different facial emotions using various types of machine learning algorithms. The inscribed circle area of the triangle (ICAT) feature gives a maximum mean classification rate of 98.17% using a Random Forest (RF) classifier compared to other features and classifiers in distinguishing emotional expressions.

Training AI-Based Feature Extraction Algorithms, for Micro CT Images, Using Synthesized Data

X-ray computed tomography (CT) is a powerful technique for non-destructive volumetric inspection of objects and is widely used for studying internal structures of a large variety of sample types. The raw data obtained through an X-ray CT practice is a gray-scale 3D array of voxels. This data must undergo a geometric feature extraction process before it can be used for interpretation purposes. Such feature extraction process is conventionally done manually, but with the ever-increasing trend of image data sizes and the interest in identifying more miniature features, automated feature extraction methods are sought. Given the fact that conventional computer-vision-based methods, which attempt to segment images into partitions using techniques such as thresholding, are often only useful for aiding the manual feature extraction process, machine-learning based algorithms are becoming popular to develop fully automated feature extraction processes. Nevertheless, the machine-learning algorithms require a huge pool of labeled data for proper training, which is often unavailable. We propose to address this shortage, through a data synthesis procedure. We will do so by fabricating miniature features, with known geometry, position and orientation on thin silicon wafer layers using a femtosecond laser machining system, followed by stacking these layers to construct a 3D object with internal features, and finally obtaining the X-ray CT image of the resulting 3D object. Given that the exact geometry, position and orientation of the fabricated features are known, the X-ray CT image is inherently labeled and is ready to be used for training the machine learning algorithms for automated feature extraction. Through several examples, we will showcase: (1) the capability of synthesizing features of arbitrary geometries and their corresponding labeled images; and (2) use of the synthesized data for training machine-learning based shape classifiers and features parameter extractors.

Local Feature Extraction Network for Point Cloud Analysis

Geometric feature extraction of 3D point clouds plays an important role in many 3D computer vision applications such as region labeling, 3D reconstruction, object segmentation, and recognition. However, hand-designed features on point clouds lack semantic information, so cannot meet these requirements. In this paper, we propose local feature extraction network (LFE-Net) which focus on extracting local feature for point clouds analysis. Such geometric features learning from a relation of local points can be used in a variety of shape analysis problems such as classification, part segmentation, and point matching. LFE-Net consists of local geometric relation (LGR) module which aims to learn a high-dimensional local feature to express the relation between points and their neighbors. Benefiting from the additional singular values of local points and hierarchical neural networks, the learned local features are robust to permutation and rigid transformation so that they can be transformed into 3D descriptors. Moreover, we embed prior spatial information of the local points into the sub-features for combining features from multiple levels. LFE-Net achieves state-of-the-art performances on standard benchmarks including ModelNet40, ShapeNetPart.

Performance Evaluation of the Data Clustering Techniques and Cluster Validity Indices for Efficient Toolpath Development for Incremental Sheet Forming

Abstract The goal of current research is to compare the data clustering techniques and cluster validity indices for geometrical feature extraction using point cloud. Here, the point clouds are generated by slicing of the computer-aided design (CAD) surface, and the data on each slice is used as inputs to the clustering algorithms. The clustering techniques are used to detect the multiple closed contours on the planer datasets. In this paper, the four most popular clustering techniques, i.e., partition-based (K-means), density-based (DBSCAN), single linkage hierarchical clustering, a variant of hierarchical clustering, and graph-based (spectral) clustering technique are compared using four different datasets. The single linkage hierarchical clustering is preferred over the other variants as it detects the arbitrarily shaped clusters efficiently and effectively. The comparison is based on the ability of successful detection of the closed contours on the planer dataset, the time required, and the input needed for the algorithm. From the investigations, it is found that DBSCAN is the most suitable technique for the feature-based toolpaths (FBTs) development. Besides, for the quality assessment of the clustering solutions and to pinpoint the superlative validity indices, techniques like Calinski-Harabasz, Ball-Hall, Davies-Bouldin, Dunn, Det Ratio, Silhouette, Trace WiB, and Log Det Ratio are compared. The solutions of the clustering techniques are used to compare the validity indices. On the basis of comparative analysis, it is concluded that the Ball-Hall, Dunn, Det Ratio, and Log Det Ratio indices are the best validity criterions for the arbitrary shaped closed contours. The overall outcomes of this research will help in building the algorithms for the feature-based toolpath development strategies for the various manufacturing processes using data science and machine learning techniques.

Efficient Feature Extraction from Multispectral Images for Face Recognition Applications: A Deep Learning Approach

In recent years many face recognition algorithms were used for the identification and authentication of a person to a system. However, still, feature extraction from multispectral images was considered to be a challenging task. Feature extraction, including highlight location and portrayal, assumes a significant job in real-time security-based applications. In this paper, a novel Geometric Algebra-based Multivariate Regression Feature Extraction (GA-MVRFE) algorithm was proposed to extract features from a huge dataset stored in the cloud efficiently. This proposed algorithm works with the supreme expedient deep learning approach - Convolutional Neural Network (CNN) for image classification. CNN will automatically detect significant features from the multispectral images without any human intrusion from a huge database. Real-time images were captured with three different cameras and applied filters over the images and were created as a dataset. To show the competence of the proposed algorithm, an exclusively created dataset with a set of 14,400 image data was applied in the proposed and other existing algorithms, and their efficiency and robustness were noted. Providentially, GA-MVRFE produced better accuracy in ‘Face Recognition’ with a less time fraction compared with former algorithms. Obtained accuracy % for Geometric Algebra Oriented fast and Rotated Brief (GA-ORB), Geometric Algebra Fast Retina key-point Extraction Algorithm (GA-FREAK), Trilateral Smooth Filtering (TRSF), Cross Regression Multiple View Features extraction (CRMVF) and GA-MVRFE was 87.81, 83.23, 90.72, 91.67 and 97.57 respectively.

Extended Geometric Feature Extraction Process for Detecting Multiple Frequency Oscillations in KEPCO System

In this study, a new algorithm dealing with time-varying modes for determining and tracing multiple frequency oscillations in power systems is proposed. Multiple modes or resonance of forced and natural oscillations can have severe effects on a power system. Therefore, it is crucial to recognize the dominant natural and forced oscillations because the oscillation mode possesses time-varying features that depend on the system operating conditions or changes in the parameters. The salient features of the proposed algorithm include the use of a time-series-based approach to recognize undesired modes (including multiple oscillations over a wide frequency range), tracing time-varying modes as the power system operating condition changes, and effectively determining and applying the oscillation features before implementing the corresponding control measures in the power system. In this study, multiple frequency oscillation scenarios for the test system and practical measurement data for a recent incident that occurred in Korea Electric Power Corporation (KEPCO) system are discussed. Therefore, the proposed algorithm can be practically applied in wide-area monitoring systems, not only for a single forced oscillation or local mode detection but also for system-wide inter-area mode recognition.

Automated coronary artery tree segmentation in coronary CTA using a multiobjective clustering and toroidal model-guided tracking method

Background and objectiveAccurate coronary artery tree segmentation can now be developed to assist radiologists in detecting coronary artery disease. In clinical medicine, the noise, low contrast, and uneven intensity of medical images along with complex shapes and vessel bifurcation structures make coronary artery segmentation challenging. In this work, we propose a multiobjective clustering and toroidal model-guided tracking method that can accurately extract coronary arteries from computed tomography angiography (CTA) imagery. MethodsUtilizing integrated noise reduction, candidate region detection, geometric feature extraction, and coronary artery tracking techniques, a new segmentation framework for 3D coronary artery trees is presented. The candidate regions are extracted using a multiobjective clustering method, and the coronary arteries are tracked by a toroidal model-guided tracking method. ResultsThe qualitative and quantitative results demonstrate the effectiveness of the presented framework, which achieves better performance than the compared segmentation methods in three widely used evaluation indices: the Dice similarity coefficient (DSC), Jaccard index and Recall across the CTA data. The proposed method can accurately identify the coronary artery tree with a mean DSC of 84%, a Jaccard index of 74%, and a Recall of 93%. ConclusionsThe proposed segmentation framework effectively segments the coronary tree from the CTA volume, which improves the accuracy of 3D vascular tree segmentation.

Mesh Denoising With Facet Graph Convolutions.

We examine the problem of mesh denoising, which consists of removing noise from corrupted 3D meshes while preserving existing geometric features. Most mesh denoising methods require a lot of mesh-specific parameter fine-tuning, to account for specific features and noise types. In recent years, data-driven methods have demonstrated their robustness and effectiveness with respect to noise and feature properties on a wide variety of geometry and image problems. Most existing mesh denoising methods still use hand-crafted features, and locally denoise facets rather than examine the mesh globally. In this work, we propose the use of a fully end-to-end learning strategy based on graph convolutions, where meaningful features are learned directly by our network. It operates on a graph of facets, directly on the existing topology of the mesh, without resampling, and follows a multi-scale design to extract geometric features at different resolution levels. Similar to most recent pipelines, given a noisy mesh, we first denoise face normals with our novel approach, then update vertex positions accordingly. Our method performs significantly better than the current state-of-the-art learning-based methods. Additionally, we show that it can be trained on noisy data, without explicit correspondence between noisy and ground-truth facets. We also propose a multi-scale denoising strategy, better suited to correct noise with a low spatial frequency.

Generalizability of Convolutional Encoder–Decoder Networks for Aerodynamic Flow-Field Prediction Across Geometric and Physical-Fluidic Variations

Abstract The generalizability of a convolutional encoder–decoder based model in predicting aerodynamic flow field across various flow regimes and geometric variation is assessed. A rich master dataset consisting of 11,000+ simulations including cambered, uncambered, thin, and thick airfoils simulated at varying angles of attack is generated. The various Mach and Reynolds number (Re) chosen allows analysis across compressible, incompressible, low, and high Re flow regimes. Multiple studies are carried out with the model trained on datasets that are categorized based on the aforementioned parameters. In each study, the loss of prediction accuracy by training the model on a larger dataset (generalizability), versus a smaller categorically sorted dataset, is evaluated. Largely disparate flow features across the Re range lead to a 25.56% loss, while the generalization across Mach range led to an average of 23.95% loss. However, flow-field changes induced due to geometric variation exhibited a better generalization potential, through an increased accuracy of 12.4%. The encoder–decoder architecture allows extraction of relevant geometric features from largely different geometries (geometric generalization) providing a better out-of-sample prediction accuracy in comparison to physics-based generalization. It is shown that, through user-informed choice of training data (removal of geometrically similar samples), computational costs incurred in generating training data can be reduced. This is important for the application of such methods in the design optimization of platforms and components that require the analysis of the fluid flows.

Projected fuzzy C-means clustering with locality preservation

Traditional partition-based clustering algorithms, hard or fuzzy version of C-means, could not deal with high-dimensional data sets effectively as redundant features may impact the computation of distances and local spatial structures among patterns are rarely considered. High dimensionality of space gives rise to so-called concentration effect that is detrimental. In this paper, a novel locality preserving based fuzzy C-means (LPFCM) clustering method and its optimization are presented. An orthogonally projected space, which preserves the locality of structural properties, can be generated in LPFCM, thus enhancing the capability of fuzzy C-means (FCM) for handling high-dimensional data. It is the first time to introduce projection techniques to the FCM optimization objective function directly, and the ideas of fuzzy clustering, geometric structure preservation and feature extraction are seamlessly integrated. LPFCM is also regarded as a unified model that combines two separate stages of spectral clustering. Experimental results on some benchmark data sets show the effectiveness of LPFCM in comparison with FCM and some state-of-the-art methods.

A deep learning system for single and overall weight estimation of melons using unmanned aerial vehicle images

Generation of yield maps enables making agronomic decisions related to resource management and marketing, leading to improved production and breeding processes. Estimating melon yield production before harvest at single-melon resolution is a labor-intensive task, requiring a detailed account of accumulated yield and general yield distribution, as well as detailed measurements of melon size and location. This study presents an algorithmic pipeline for detection and yield estimation of melons from top-view color images acquired by a digital camera mounted on an unmanned aerial vehicle. The yield estimation provides both the number of melons and the weight of each melon. The system includes three main stages: melon detection, geometric feature extraction, and individual melon yield estimation. The melon-detection process was based on the RetinaNet deep convolutional neural network. Transfer learning was used for the training to detect small objects in high-resolution images successfully. The detection process achieved an average precision score of 0.92 with a F1 score of more than 0.9 in a variety of agricultural environments. For each detected melon, feature extraction was applied using the Chan–Vese active contour algorithm and principal component analysis ellipse-fitting method. A regression model that ties the ellipse features to the melon’s weight is presented. The modified (adjusted) RAdj2 value of the regression model was 0.94. The system results for estimating the weight of a single melon measured by the mean absolute percentage error index achieved 16%. The analysis revealed that this could be decreased to 12% error with more accurate geometrical feature extraction. Overall yield estimation derived by summing the weights of all melons in the field resulted in only a 3% underestimation of the actual total yield.

Fast adaptive fuzzy enhancement and correlation features analysis of flame image of sintering section

The state of the sintering end point can be indirectly reflected by the flame image characteristics of the material layer section at the end of the sintering machine. However, the image of tail section collected by industrial camera is easy to be interfered by smoke, dust and thermal radiation. As a result, the edge between the flame area and the material layer area becomes fuzzy, accompanied with halo and noise, which leads to the degradation of flame image. In order to solve the problem of image quality degradation, a new method based on weighted guided image filtering and fast adaptive fuzzy enhancement of flame image of sintering cross section is proposed in this paper; furthermore, the correlation analysis of the flame image characteristics of sintering section is carried out. The main contents of this paper include three parts: cross-sectional flame image enhancement, image brightness characteristics and geometric feature extraction, and image feature correlation analysis. The results show that the proposed method effectively eliminates the interference of noise and halo in the cross-sectional flame image. The brightness characteristics of the flame image are related to the length and height of the flame and the area of the red fire region, while there is no correlation between the brightness characteristics of the flame image and the centroid variance. Therefore, the brightness characteristics and the centroid variance can be used as the input feature for the discrimination of sintering state.

Wavelet features embedded convolutional neural network for multiscale ear recognition

The ear has emerged as a new biometric trait to recognize humans from their profile faces. Stability over the years, noninvasive capturing process, expressionless images, and significant variation in shape among individuals make the ear a suitable choice when compared with other biometrics. Convolutional neural network (CNN)’s capability to learn and discriminate specific features irrespective of image variation makes it the most suitable choice for ear recognition in comparison to local geometrical feature extraction methods. The recognition performance affecting factors such as occlusion, image resolution, and head rotation during ear acquisition in an uncontrolled environment makes it a challenging task, and the use of spatial features only in CNN limits the recognition rate. We explored the various CNN configurations and propose the spectral–spatial feature-based CNN. An embedding algorithm is proposed to fuse multilevel spectral information from the image domain with spatial features extracted at each deep layer of the CNN network. The poor convergence problem of CNN is tackled by introducing batch normalization after every convolution. The analysis of the result shows that the proposed framework is superior among all CNN achieving 98.15% and 78.88% accuracy on the Annotated Web Ears with 1000 images and earVN1.0 with 28,412 images, respectively. In addition to fine-tuning and configuring the layers of CNN, the fusion of missing spectral features and spatial features can strengthen the CNN.