Geometric Filter Research Articles

Tightly coupled stereo vision-inertial odometry based on point and line feature

Abstract To improve the accuracy and robustness of visual simultaneous localization and mapping (SLAM) in low-texture environments, this paper proposes a robust and fast stereo vision inertial SLAM pose estimation method that combines point and line features with an inertial measurement unit (IMU). The method tightly couples visual point and line features with IMU constraints, forming a least-squares problem through the error of each constraint term for nonlinear optimization. To address the issues of over-segmentation and time consumption in traditional line segment detection (LSD) algorithms, an improved LSD algorithm is adopted to accelerate line feature extraction. This approach merges nearby line segments based on spatial geometric relationships and filters out invalid segments, improving the time efficiency of the algorithm. Finally, experiments conducted in low-texture environments demonstrate that our algorithm achieves high localization accuracy and robustness.

Exploring Topological Information Beyond Persistent Homology to Detect Geospatial Objects

Accurate detection of geospatial objects, particularly landslides, is a critical challenge in geospatial data analysis due to the complex nature of the data and the significant consequences of these events. This paper introduces an innovative topological knowledge-based (Topological KB) method that leverages the integration of topological, geometrical, and contextual information to enhance the precision of landslide detection. Topology, a fundamental branch of mathematics, explores the properties of space that are preserved under continuous transformations and focuses on the qualitative aspects of space, studying features like connectivity and exitance of loops/holes. We employed persistent homology (PH) to derive candidate polygons and applied three distinct strategies for landslide detection: without any filters, with geometrical and contextual filters, and a combination of topological with geometrical and contextual filters. Our method was rigorously tested across five different study areas. The experimental results revealed that geometrical and contextual filters significantly improved detection accuracy, with the highest F1 scores achieved when employing these filters on candidate polygons derived from PH. Contrary to our initial hypothesis, the addition of topological information to the detection process did not yield a notable increase in accuracy, suggesting that the initial topological features extracted through PH suffices for accurate landslide characterization. This study advances the field of geospatial object detection by demonstrating the effectiveness of combining geometrical and contextual information and provides a robust framework for accurately mapping landslide susceptibility.

Minimum-energy switching geometric filter on lie groups for differential-drive wheeled mobile robots

Accurate state estimation plays a critical role in various applications, such as tracking, regulation, and fault detection in robotic and mechanical systems. Typically, the Kalman–Bucy filter is used as a linear state observer for this purpose. However, real-world robots often exhibit complex behavior, characterized by a combination of dynamics, making it essential to employ hybrid filters. In this context, the Switching Kalman filter stands out as a well-established solution. In this article we aim to generalize the Brownian-Markov Stochastic Model, a hybrid dynamic model for differential-drive wheeled mobile robots, to the case of a mobile robot whose center of mass is not aligned to the wheels axle middle point, and to design a geometric hybrid state estimator by exploiting the Lie groups theory. The Brownian-Markov Stochastic Model features two modes: “grip” and “slip”. These modes correspond to ideal grip and lateral slippage, with transitions governed by a state-dependent Markov chain. To validate the proposed switching filter, we conduct MATLAB® simulations of the robot’s motion in a scenario prone to lateral grip loss, comparing the state estimates produced by the switching geometric filter with those obtained using the switching Kalman filter.

Localization of asparagus spears using time-of-flight imaging for robotic harvesting

PurposeTime of flight (ToF) imaging is a promising emerging technology for the purposes of crop identification. This paper aim to presents localization system for identifying and localizing asparagus in the field based on point clouds from ToF imaging. Since the semantics are not included in the point cloud, it contains the geometric information of other objects such as stones and weeds other than asparagus spears. An approach is required for extracting the spear information so that a robotic system can be used for harvesting.Design/methodology/approachA real-time convolutional neural network (CNN)-based method is used for filtering the point cloud generated by a ToF camera, allowing subsequent processing methods to operate over smaller and more information-dense data sets, resulting in reduced processing time. The segmented point cloud can then be split into clusters of points representing each individual spear. Geometric filters are developed to eliminate the non-asparagus points in each cluster so that each spear can be modelled and localized. The spear information can then be used for harvesting decisions.FindingsThe localization system is integrated into a robotic harvesting prototype system. Several field trials have been conducted with satisfactory performance. The identification of a spear from the point cloud is the key to successful localization. Segmentation and clustering points into individual spears are two major failures for future improvements.Originality/valueMost crop localizations in agricultural robotic applications using ToF imaging technology are implemented in a very controlled environment, such as a greenhouse. The target crop and the robotic system are stationary during the localization process. The novel proposed method for asparagus localization has been tested in outdoor farms and integrated with a robotic harvesting platform. Asparagus detection and localization are achieved in real time on a continuously moving robotic platform in a cluttered and unstructured environment.

Geometric Graph Filters and Neural Networks: Limit Properties and Discriminability Trade-Offs

Geometric Graph Filters and Neural Networks: Limit Properties and Discriminability Trade-Offs

On Skin Model Describability using Laplace-Beltrami Operator for Geometrical Product Specification and Verification

The skin model serves as a foundational concept defined in ISO standards on Geometrical Product Specifications and verification (ISO GPS), facilitating the consideration of geometric variations on the surfaces of mechanical products. However, the lack of a solid scientific foundation and practical operational applications has become a barrier for skin model concept to keep up with the development of standardized feature operations. A previous investigation was devoted to a scientific approach to defining and interpreting the skin model for ISO GPS, grounded in the structure of solids and set theory. In this paper, the use of the Laplace-Beltrami Operator (LBO) is explored as a mathematical operation to enhance the computational representations of the skin model. The describability of skin model is achieved by applying LBO in both continuous and discrete cases. Geometric filter banks, established through LBO, are employed for skin model to identify defects at various scales of observation. An application of LBO is demonstrated through a case study to highlight the use of established geometric filter banks for skin model. Further discussions show that the spectral analysis of a surface using LBO enables the characterization of skin model, contributing to the current and future standardization of geometrical product specification and verification. Finally, conclusion and future works are highlighted.

Approaches to X-ray CT Evaluation of In-Situ Experiments on Damage Evolution in an Interpenetrating Metal-Ceramic Composite with Residual Porosity

An interpenetrating metal-ceramic composite of AlSi10Mg and an open porous alumina foam, with residual porosity is investigated for the material damage under compressive load within an X-ray CT in-situ load stage. The focus of the research is on damage detection and evaluation with the commercial Avizo® software by ThermoFisher Scientific. Four different approaches are used to detect the material damage and compared afterward on their efficiency in detecting the material damage volume but not the porosity within the material. Image Stack Processing combined with different filtering techniques, as well as Digital Volume Correlation is used in this work to separate the material porosity and the material damage. For the here investigated material system with mainly spherical pores, a geometrical filter was very successful to separate porosity and damage. Nevertheless, the Digital Volume Correlation based approach showed many advantages in damage detection and turned out to be the approach of choice regarding damage onset.

Multisource forest point cloud registration with semantic-guided keypoints and robust RANSAC mechanisms

• The proposed method was independent of single tree detection and ground filtering. • A new type of keypoint was proposed for forest point cloud registration. • Robust RANSAC mechanism was introduced to enhance registration performance. The Increasing availabilities of Terrestrial Laser Scanning (TLS) and Unmanned Aerial Vehicle Laser Scanning (ULS) have advanced accurate and detailed forest structural measurements. Registration of multi-perspective observations from different platforms is a prerequisite for a comprehensive forest structure understanding. Currently, forest point cloud registrations are typically done by single tree attribute-based (e.g. tree location, stem diameter) methods, which suffer from various and complex forest compositions and terrains, and can be unreliable for forests with regular tree layouts and insufficient common trees. Therefore, this study presents a unified and marker-free framework for both TLS-TLS and ULS-TLS point cloud registrations in forest areas. First, motivated by wood-leaf separation, semantic-guided keypoints from TLS and ULS point clouds are detected by utilizing a wood response indicator (WRI), which represents reliable repeatability between different views of forest point clouds. Second, an initial correspondence set is generated by using the WRI filter and Binary Shape Context (BSC) descriptor matching. Finally, the initial correspondence set is trimmed and optimized through a robust RANSAC mechanism, consisting of a geometric compatibility filter to guide and trim the hypothesis generation process and a modified hypothesis evaluation step to further optimize the transformation. Experiments have been conducted on six plots involving diverse stem densities and tree species located in forest farms in Guangxi, China. The resulting average registration residual and runtime are 0.049 m and 93 s for TLS-TLS scenes, 0.299 m and 242 s for ULS-TLS scenes, respectively. Comprehensive comparisons demonstrate that the proposed method outperforms other baselines. It is demonstrated that the proposed framework can improve the practicality and efficiency of multi-source data collection and registration, thus facilitating the application of TLS and ULS in forest ecosystem science.

Development of Machine Learning and Medical Enabled Multimodal for Segmentation and Classification of Brain Tumor Using MRI Images

The improper and excessive growth of brain cells may lead to the formation of a brain tumor. Brain tumors are the major cause of death from cancer. As a direct consequence of this, it is becoming more challenging to identify a treatment that is effective for a specific kind of brain tumor. The brain may be imaged in three dimensions using a standard MRI scan. Its primary function is to examine, identify, diagnose, and classify a variety of neurological conditions. Radiation therapy is employed in the treatment of tumors, and MRI segmentation is used to guide treatment. Because of this, we are able to assess whether or not a piece that was spotted by an MRI is a tumor. Using MRI scans, this study proposes a machine learning and medically assisted multimodal approach to segmenting and classifying brain tumors. MRI pictures contain noise. The geometric mean filter is utilized during picture preprocessing to facilitate the removal of noise. Fuzzy c-means algorithms are responsible for segmenting an image into smaller parts. The identification of a region of interest is facilitated by segmentation. The GLCM Grey-level co-occurrence matrix is utilized in order to carry out the process of dimension reduction. The GLCM algorithm is used to extract features from photographs. The photos are then categorized using various machine learning methods, including SVM, RBF, ANN, and AdaBoost. The performance of the SVM RBF algorithm is superior when it comes to the classification and detection of brain tumors.

Lung Cancer Classification and Prediction Using Machine Learning and Image Processing.

Lung cancer is a potentially lethal illness. Cancer detection continues to be a challenge for medical professionals. The true cause of cancer and its complete treatment have still not been discovered. Cancer that is caught early enough can be treated. Image processing methods such as noise reduction, feature extraction, identification of damaged regions, and maybe a comparison with data on the medical history of lung cancer are used to locate portions of the lung that have been impacted by cancer. This research shows an accurate classification and prediction of lung cancer using technology that is enabled by machine learning and image processing. To begin, photos need to be gathered. In the experimental investigation, 83 CT scans from 70 distinct patients were utilized as the dataset. The geometric mean filter is used during picture preprocessing. As a consequence, image quality is enhanced. The K-means technique is then used to segment the images. The part of the image may be found using this segmentation. Then, classification methods using machine learning are used. For the classification, ANN, KNN, and RF are some of the machine learning techniques that were used. It is found that the ANN model is producing more accurate results for predicting lung cancer.

An enhanced RSS-distance-angle weighted geometric filter for device-free localization

Device-free localization and tracking as an emerging technology has attracted substantial research attention in wireless sensor networks. However, there is much room for improvement in localization and tracking accuracy. In this paper, by using received signal strength (RSS) measurements we propose an enhanced geometric filter which consists of a series of weights corresponding to RSS, distance, and angle, respectively. Specifically, RSS-based weights are dependent on the change in RSS of communication links to remove improbable target locations. To ensure robust tracking performance, distance-based weights are assigned to probable target locations. Angle-based weights are used to ensure the correct direction of motion. Experimental results demonstrate that the proposed geometric filter could improve the accuracy of positioning by up to 50.8% for non-straight path and 48.3% for straight path over some state-of-the-art methods.

Fully Automated Algorithm for Light Pole Detection and Mapping in Rural Highway Environment Using Mobile Light Detection and Ranging Point Clouds

The widespread use of light detection and ranging (LiDAR) data provides a promising source for the automatic detection and inventory of roadside assets. One of the essential elements in roadside furniture is light poles. There is limited research on the mapping of light poles using point-cloud data on rural highways. In this environment, the placement of light poles within roadside clear zones often poses a safety concern, as they are related to an increased risk of collisions. Only a limited number of studies have explored the relationship between light poles and safety because of the time-consuming and labor-intensive practices of collecting light pole assets data using traditional manual methods. This paper proposes an automated approach to mapping the locations of light poles. First, the scanning vehicle trajectory is extracted, smoothed, and then used to segment the point-cloud data into smaller overlapped batches of data. Several filters are applied to extract pole-like objects from the data. The segments are combined back together, and a density-based clustering algorithm is used to group the remaining points into clusters. A geometric filter is finally applied to extract light poles. The model is tested on 28 km of data on three rural highways in Alberta, Canada. The proposed algorithm is found to be accurate relative to previous studies, with average precision, recall, and F1 scores exceeding 98% for the test segments. The proposed work can assist in the automation of light pole inventory and road safety audits by transportation agencies.

Computational Technique Based on Machine Learning and Image Processing for Medical Image Analysis of Breast Cancer Diagnosis

Breast cancer is the most lethal type of cancer for all women worldwide. At the moment, there are no effective techniques for preventing or curing breast cancer, as the source of the disease is unclear. Early diagnosis is a highly successful means of detecting and managing breast cancer, and early identification may result in a greater likelihood of complete recovery. Mammography is the most effective method of detecting breast cancer early. Additionally, this instrument enables the detection of additional illnesses and may provide information about the nature of cancer, such as benign, malignant, or normal. This article discusses an evolutionary approach for classifying and detecting breast cancer that is based on machine learning and image processing. This model combines image preprocessing, feature extraction, feature selection, and machine learning techniques to aid in the classification and identification of skin diseases. To enhance the image’s quality, a geometric mean filter is used. AlexNet is used for extracting features. Feature selection is performed using the relief algorithm. For disease categorization and detection, the model makes use of the machine learning techniques such as least square support vector machine, KNN, random forest, and Naïve Bayes. The experimental investigation makes use of MIAS data collection. This proposed technology is advantageous for accurately identifying breast cancer disease using image analysis.

Study on the Effect of Damping Asymmetry of the Vertical Suspension on the Railway Bogie Vibrations

This paper has an original approach to the study of vibrations in the bogie of the railway vehicle by taking into account the asymmetry of the vertical damping of suspension. The study relies on the results derived from the numerical simulation applications, where the damping asymmetry of the bogie suspension is simulated through different degrees of reducing the damping of the suspension of one of the wheels compared to the reference value. Under such circumstances, the vibrations behaviour of the bogie, assessed in five reference chassis points—four points located against the suspension of each wheel and one point located against the bogie’s center of mass—will be the result of several combined effects. A first effect is produced by the suspension asymmetry that trigger interferences between the vertical vibrations of bounce and pitch and the horizontal roll vibrations of the bogie. Another effect is introduced by the reduction of the damping in the system. To these, the geometric filtering effect due to the bogie wheelbase is added, which has a selective nature depending on velocity. The analysis of the bogie vertical vibrations aims to assess the level of vibrations and identify the dominant vibration mode in the bogie reference points, as well as to determine the critical point of the vibration mode of the bogie, in correlation with velocity and the reduction degree of the suspension damping.

Efficient image-based indoor localization with MEMS aid on the mobile device

Recently, imaged-based localization has become a hotspot in indoor localization research due to its high accuracy and no additional infrastructure requirements. However, computational complexity and memory consumption are critical issues for terminal devices in natural large-scale indoor scenes. In this paper, we propose a system that takes advantage of built-in sensors, a microelectromechanical system (MEMS), and 3D model-based localization to overcome resource constraints. Furthermore, we propose a rough carrier six-degree-of-freedom (6-DoF) pose estimation and error analysis based on the pedestrian dead reckoning and attitude and heading reference systems (PDR/AHRS) with pedestrian motion constraints. We additionally recommend a geometric pose error propagation model to compute the 3D point pose uncertainty of corresponding 2D features on the query images and a matching pool to extract the activated points for determining the 2D-3D correspondence, which efficiently rejects the outlier through the geometric filter in the RANSAC loop of pose estimation. Moreover, based on the fusion of results from the MEMS-based and image-based localization, we can obtain an optimized carrier pose by using a dynamic reliability strategy to ensure the continuous robustness of the localization system and avoid failures when the matching lacks sufficient inliers. To evaluate the performance of the system, we conducted extensive experiments in large-scale indoor scenes. The results demonstrate that our image-based localization system generates excellent 2D-3D correspondence with the MEMS aid and that the proposed strategy obtains accurate 6-DoF poses in almost real time by using a smartphone. When compared to the conventional straightforward 2D-3D method (ACS) for localization accuracy and usability tests in various indoor scenes, the proposed method can save time by more than 80% and reduce translation and rotation errors by 27% and 10%, respectively. We also show that our new localization system has robust performance and stability for various indoor scenes and influences by different image resolutions and multiview gestures. Furthermore, our idea has a good application foreground, as 3D indoor data acquisition and pose estimation algorithms are compatible with the current state of data expansion for pedestrian navigation.

Robust State Estimation With Maximum Correntropy Rotating Geometric Unscented Kalman Filter

The geometric unscented Kalman filter (GUF) is an effective method for state estimation due to its advantages of high precision, reasonable efficiency, and good stability. In order to further improve the filtering performance of GUF, multiple rotation matrices are first designed by optimizing the rotation angles of their corresponding sampling points with maximum likelihood estimation, leading to a novel sampling strategy, namely, rotating geometric unscented sampling (RGUS). Then, applying RGUS to the GUF framework generates a rotating geometric unscented Kalman filter (RGUF). Since observable measurements are generally corrupted by non-Gaussian noise, RGUF may suffer from performance degradation due to the used minimum mean square error (MMSE) criterion. To improve the robustness of RGUF against non-Gaussian noises, we propose a novel maximum correntropy rotating geometric unscented Kalman filter (MCRGUF) using the maximum correntropy criterion (MCC). Finally, a Cramér–Rao lower bound (CRLB) of MCRGUF is introduced as a performance indicator. Simulations on three examples validate the high filtering accuracy and strong robustness of MCRGUF in the presence of non-Gaussian noise.

A Class of Nonlinear Kalman Filters Under a Generalized Measurement Model With False Data Injection Attacks

In this letter, a generalized measurement model (GMM) is established under the attack injecting additive and multiplicative false data, simultaneously. The GMM can be applied to any existing nonlinear Kalman filters (NKFs), such as extended Kalman filter (EKF), cubature Kalman filter (CKF), and geometric unscented Kalman filter (GUF). Based on the constructed GMM, a class of nonlinear Kalman filters with GMM (NKFs-GMM) is proposed to deal with additive false data, multiplicative false data, and simultaneous attacks on additive and multiplicative false data with no increase of computational complexity. Numerical simulations show that the filtering accuracy of NKFs-GMM is superior to that of corresponding NKFs.

Region-Based Polarimetric Covariance Difference Matrix for PolSAR Ship Detection

To more effectively detect small ships, in this article, a novel region-based polarimetric covariance difference matrix [RP] is put forward, which mainly consists of two stages. Briefly speaking, in the first stage, a new pixel representation way is proposed to depict the spatial characteristics of pixel, through which the difference information related to pixel’s local region is calculated as well. In the second stage, the global region difference information of pixel is computed. Finally, we construct [RP] via fusing these two different kinds of information together with a balance factor <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$c$ </tex-math></inline-formula> . Meanwhile, considering that the backscattering energy of ships is useful for ship detection, a new intensity-driven polarimetric notch filter (ID-PNF <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">RP</sub> ) is also derived from [RP]. Three different datasets are adopted to evaluate the effectiveness of [RP] and ID-PNF <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">RP</sub> . Experimental results show that: 1) compared with the polarimetric covariance matrix [ <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$C$ </tex-math></inline-formula> ] and the polarimetric covariance difference matrix [ <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$P$ </tex-math></inline-formula> ], [RP] is more suitable for ship detection and 2) compared with the original geometrical perturbation-polarimetric notch filter (GP-PNF) and the total power detector SPAN, the proposed method ID-PNF <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">RP</sub> can better detect small ships with greater figure of merit (FoM) and target-to-clutter ratio (TCR) values.

Geometric Midpoint Algorithm for Device-Free Localization in Low-Density Wireless Sensor Networks

Device-free localization (DFL) is a technique used to track a target transporting no electronic devices. Radiofrequency (RF) tomography based DFL technology in wireless sensor networks has been a popular research topic in recent years. Typically, high-tracking accuracy requires a high-density wireless network which limits its application in some resource-limited scenarios. To solve this problem, a geometric midpoint (GM) algorithm based on the computations of simple geometric objects is proposed to realize effective tracking of moving targets in low-density wireless networks. First, we proposed a signal processing method for raw RSS signals collected from wireless links that can detect the fluctuations caused by a moving target effectively. Second, a geometric midpoint algorithm is proposed to estimate the location of the target. Finally, simulations and experiments were performed to validate the proposed scheme. The experimental results show that the proposed GM algorithm outperforms the geometric filter algorithm, which is a state-of-the-art DFL method that yields tracking root-mean-square errors up to 0.86 m and improvements in tracking accuracy up to 67.66%.

Investigation of crackling noise in the vibration isolation systems of the KAGRA gravitational wave detector

Gravitational wave detectors, such as KAGRA, require complex mirror suspension systems to reach high sensitivity. One particular concern in these suspensions is the presence of crackling noise, where motion of the steel crystal structure disturbs material defects, causing impulse force release and exciting mirror motion. Crackling in the Geometric Anti Spring filters (GAS) in KAGRA's suspension system could introduce noise in the gravitational wave detection band. We investigate crackling noise in a miniature GAS with a low-frequency cyclical stress. We developed a scaling law between miniature and KAGRA GAS. Applying experimental results to the scaling law allows us to estimate the upper limit of crackling noise in KAGRA. It is found that crackling noise should be below the target sensitivity for gravitational waves above 60 Hz.