Automatic Filtering Research Articles

'TLDetect': AI-Based Application for Detection and Correction of Anomalous TLD Glow Curves.

This research reviews a novel artificial intelligence (AI)-based application called TLDetect, which filters and classifies anomalous glow curves (GCs) of thermoluminescent dosimeters (TLDs). Until recently, GC review and correction in the lab were performed using an old in-house software, which uses the Microsoft Access database and allows the laboratory technician to manually review and correct almost all GCs without any filtering. The newly developed application TLDetect uses a modern SQL database and filters out only the necessary GCs for technician review. TLDetect first uses an artificial neural network (ANN) model to filter out all regular GCs. Afterwards, it automatically classifies the rest of the GCs into five different anomaly classes. These five classes are defined by the typical patterns of GCs, i.e., high noise at either low or high temperature channels, untypical GC width (either wide or narrow), shifted GCs whether to the low or to the high temperatures, spikes, and a last class that contains all other unclassified anomalies. By this automatic filtering and classification, the algorithm substantially reduces the amount of the technician's time spent reviewing the GCs and makes the external dosimetry laboratory dose assessment process more repeatable, more accurate, and faster. Moreover, a database of the class anomalies distribution over time of GCs is saved along with all their relevant statistics, which can later assist with preliminary diagnosis of TLD reader hardware issues.

Reference-then-supervision framework for infrared and visible image fusion

Infrared and visible image fusion has drawn increasing attention of researchers in recent years, wherein the complementary information between two source images is extracted to synthesize a new fusion image with richer information. Deep neural network is the latest technique to tackle various image fusion problems. However, the natural absence of ground truth makes it very challenging to optimize deep learning models. While the current methods mainly consider to uses the two source images themselves or their visual features, to provide supervision for learning, which is easy to result in the imbalance between detail preservation and brightness distribution. To boost the performance of the fusion models, it is meaningful and necessary to establish a flexible framework that can combine the advantages of existing models to produce reliable supervision for model training. In this work, we propose a novel reference-then-supervision framework, which aims to fully leverage and exploit the available favorable reference information based on the performance of existing methods and then construct high-quality reliable supervision to assist in model building. For this purpose, we design an automatic filter to produce favorable reference and devise an adaptive enhancement method to construct reliable supervision, which helps to aggregate the advantages of various existing fusion methods for yielding visually pleasing results adapting to different complex scenarios. Extensive experiments on two commonly used datasets and our built challenging test set demonstrate that our framework can greatly improve the performance of existing fusion methods. Ablation study and empirical analysis also present the efficacy of our framework design. Furthermore, the applications on downstream pedestrian detection and object tracking tasks indicate the great potential of our framework. Our code and data are publicly available at https://github.com/zhenglab/ReferenceSupervisionIVIF.

An FDA-Validated, Self-Cleaning Liquid Chromatography-Mass Spectrometry System for Determining Small-Molecule Drugs and Metabolites in Organoid/Organ-on-Chip Medium.

As organoids and organ-on-chip (OoC) systems move toward preclinical and clinical applications, there is an increased need for method validation. Using a liquid chromatography-mass spectrometry (LC-MS)-based approach, we developed a method for measuring small-molecule drugs and metabolites in the cell medium directly sampled from liver organoids/OoC systems. The LC-MS setup was coupled to an automatic filtration and filter flush system with online solid-phase extraction (SPE), allowing for robust and automated sample cleanup/analysis. For the matrix, rich in, e.g., protein, salts, and amino acids, no preinjection sample preparation steps (protein precipitation, SPE, etc.) were necessary. The approach was demonstrated with tolbutamide and its liver metabolite, 4-hydroxytolbutamide (4HT). The method was validated for analysis of cell media of human stem cell-derived liver organoids cultured in static conditions and on a microfluidic platform according to Food and Drug Administration (FDA) guidelines with regards to selectivity, matrix effects, accuracy, precision, etc. The system allows for hundreds of injections without replacing chromatography hardware. In summary, drug/metabolite analysis of organoids/OoCs can be performed robustly with minimal sample preparation.

An Ensemble Keyword Extraction Model for News Texts with Statistical and Graphical Features

Keyword extraction is an essential tool for many text mining applications such as automatic indexing, summarizing, classification, clustering and automatic filtering. Automated keyword extraction is essential as the daily text data to be reached and processed have increased tremendously over the Internet, e.g. millions of news articles are published daily online. In this paper, a novel ensemble model for automatic extraction of keywords from news articles is proposed. The proposed model handles keyword extraction as a sequence labeling task. Two sub-modules representing the statistical and graphical features by their calculated scores for each input token were combined in the token classification module. The Ensemble Token Classification module was trained and tested separately with the ensemble algorithms Random Forest, XgBoost, Decision Tree and Voting Classification. For training, we collected two news datasets from Kazakh and Russian news sites published in Cyrillic alphabet. We also collected an Arabic news dataset, ArabianNews. The performance of the model was also compared with the widely used 500N-KPCrowd dataset in the literature, which consists of English news content in Latin alphabet. The proposed model achieved the best performance with an [Formula: see text]-score of 0.71 and 0.86 on the 500N-KPCrowd and Russian datasets, respectively. We attained the best [Formula: see text]-score (0.97) with the KazakhNews and ArabianNews datasets.

USE OF ARTIFICIAL INTELLIGENCE METHODS AND MODELS FOR IMPROVING EXPERT SYSTEMS OF INTRUSION DETECTION

In the domain of cyber security, the efficacy of Intrusion Detection Systems (IDS) is critical for the proactive identification and mitigation of cyber threats. This research delineates a novel paradigm for enhancing IDS accuracy through the integration of advanced Artificial Intelligence (AI) methodologies, thereby setting a new benchmark in network security defense mechanisms. Utilizing a synergistic approach that combines both descriptive and inferential statistical analyses, this study introduces an expert system endowed with the capability to detect network intrusions with an unparalleled accuracy rate of 99.98%. By incorporating Extreme Gradient Boosting (XGBoost) for the classification of predefined attack vectors and a Variational Autoencoder (VAE) for anomaly detection, the research extends the boundaries of current cyber threat detection frameworks. These methodologies not only enhance the precision of threat categorization but also introduce a mechanism for the system to adapt to novel, previously unidentified cyber threats through real-time learning and adaptation to emerging data patterns. Critically, the expert system is engineered to facilitate high-speed data processing and supports online learning, making it optimally suited for application in high-traffic network environments. The scientific novelty of this research is encapsulated in the formulation of advanced AI-driven models for the dual purposes of traffic anomaly detection and the classification of cyber-attack types based on distinctive behavioral characteristics. These models are meticulously designed to evolve, learning from new data in real time, thereby continuously enhancing the system's efficacy. In practical terms, the system provides a robust solution for the protection of digital ecosystems against intrusions, enabling the automatic filtration of malicious network traffic. Beyond its immediate applicability, the study contributes to the field of cyber security by laying down a foundational framework for the future exploration of AI-based security solutions. It invites further scientific inquiry into the development of adaptive, intelligent IDS mechanisms, potentially revolutionizing the approach to cyber defense strategies.

Solutions of Surface-Wave Dispersion and Attenuation in Stratified Viscoelastic Media Using a Spectral-Element Approach

ABSTRACT Efficient and accurate calculation for the dispersion and attenuation of the surface waves in viscoelastic media is numerically challenging because the eigen wavenumbers are located in the complex domain. In this study, we propose a semianalytical spectral-element method (SASEM), which can determine the complex eigen wavenumbers by solving linear eigenvalue problems. By simplifying the structure of the eigenvalue problem, we significantly improve the calculation efficiency. The implementation of the frequency-dependent automatic discretization, semi-infinite element, and mode filter guarantees the correctness and accuracy of the modal solutions. Because no root-searching schemes are required, the root-skipping problem is naturally avoided. The numerical tests show that the SASEM can provide sufficiently accurate solutions with much less computation cost than traditional Muller’s method. Meanwhile, SASEM exhibits high flexibility when applied to media the parameters for which vary continuously with depth. To demonstrate the effectiveness of SASEM for complicated dispersion features, the dispersion curves and eigen wavefields of the viscoelastic media with a low-velocity layer are also analyzed. The results of numerical tests indicate the versatility, efficiency, and accuracy of our method. With further study, the proposed SASEM has the potential to become a promising tool for the investigation and retrieval of viscoelastic subsurface structures.

Reflectance imaging spectroscopy for the study of archaeological pre-Columbian textiles

In this study, two pre-Columbian textiles belonging to the Middle Orizon and Necropolis Paracas Cultures are studied by means of reflectance imaging spectroscopy. The non-invasive analysis of artifacts is a fundamental step when investigating the constituent materials and their state of conservation in order to target subsequent sampling campaigns in specific areas. The paper aims to explore the spectral properties in the Visible Near-Infrared (VNIR) and Short-Wave Infrared (SWIR) intervals and investigate the spectral correlation between the relevant identified regions of interest. Two ways of selecting relevant spectral signatures are proposed: manual selection and automatic filtering. According to these two selections, the regions of interest undergo spectral unmixing in the VNIR interval to investigate which dyes had been used as such. The hyperspectral images are then mapped according to the selected endmembers in both VNIR and SWIR intervals. Small areas that feature the usage of a peculiar black thread have been given particular attention and imaged with a macro objective. The SWIR mapping on one of the samples has helped to reveal a pattern that can be ascribed to the manufacturing processes or the burial conditions.

Semi-automated filter micro-solid phase extraction based on urea-formaldehyde microsphere followed by corona discharge-ion mobility spectrometry for simultaneous detection of fentanyl and nor-fentanyl in urine

In this work, semi-automated filter micro-solid phase extraction (SA-FμSPE) packed by urea–formaldehyde (UF) microspheres coupled with corona discharge ionization-ion mobility spectrometry (CD-IMS) was developed for the extraction and determination of fentanyl and nor-fentanyl in urine samples. UF microspheres, synthesized by aqueous polymerization, can be used to achieve effective extraction of analytes through multiple adsorption interactions such as hydrophobic, dipole–dipole and hydrogen bonding interactions. UF microspheres dispersed in methanol were packed in the filter by syringe, and the extraction process was performed by a portable automatic filter extraction device. The parameters affecting extraction efficiency (i.e., extraction time, the flow rate of extraction, pH value, amount of sorbent, ionic strength, cycle times of desorption, and type of eluent) were investigated. Then, the eluate containing fentanyl and nor-fentanyl was simultaneously analyzed by CD-IMS in positive mode. Under the optimal extraction conditions, the detection limits for the target compounds were 5.00 ng mL−1 and 8.00 ng mL−1, respectively. The spiked recoveries were in the range of 81.97 % −112.37 % by testing the real samples and validated by UHPLC-MS. Thus, the satisfactory results reveal that SA-FμSPE coupled with CD-IMS has great potential for on-site extraction and detection of fentanyl and nor-fentanyl in urine samples.

Automatic filter selection for Hilbert autocorrelation and cyclic repetition frequency based fPCG analysis

Background and ObjectivesFetal phonocardiography (fPCG) is a passive and non-invasive technique for measuring fetal heart rate. Due to the presence of external and internal noises, filtering is mandatory to obtain clinically usable signals from fPCG recordings. Frequency characteristics of the fPCG signal recorded from the mother's abdomen change with fetal parameters like gestation week, fetal position, etc. This makes the choice of filter frequencies challenging and unproductive. Therefore, our aim is to develop automatic filter selection based on fetal heart sounds characteristics. MethodsThis work presents algorithms for the automatic filter selection based on fPCG signal characteristics. The automatic filter selection algorithms are coupled with fetal heart rate estimation algorithms based on Hilbert-autocorrelation and cyclic repetition frequency. Three performance measures are determined to test the algorithms: accuracy, reliability, and success rate. ResultsThe automatic filter selection show significant improvement in success rate when compared with manually selected fixed frequency band filters. The success rate improves from 79% to 93% for Hilbert-autocorrelation based algorithm and 71% to 90% for cyclic repetition frequency-based algorithm. ConclusionAutomated filter selection could enable robust automated estimation of fetal heart rate from fPCG recordings across variety of clinical settings.

Quantifying and Reducing the Operator Effect in LSPIV Discharge Measurements

AbstractOperator choices, both in acquiring the video and data and in processing them, can be a prominent source of error in image‐based velocimetry methods applied to river discharge measurements. The Large Scale Particle Image Velocimetry (LSPIV) is known to be sensitive to the parameters and computation choices set by the user, but no systematic comparisons with discharge references or intercomparisons have been conducted yet to evaluate this operator effect in LSPIV. In this paper, an analysis of a video gauging intercomparison, the Video Globe Challenge 2020, is proposed to evaluate such operator effect. The analysis is based on the gauging reports of the 15 to 23 participants using the Fudaa‐LSPIV software and intents to identify the most sensitive parameters for the eight videos. The analysis highlighted the significant impact of the time interval, the grid points and the filters on the LSPIV discharge measurements. These parameters are often inter‐dependent and should be correctly set together to strongly reduce the discharge errors. Based on the results, several automated tools were proposed to reduce the operator effect. These tools consist of several parameter assistants to automatically set the orthorectification resolution, the grid and the time interval, and of a sequence of systematic and automatic filters to ensure reliable velocity measurements used for discharge estimation. The application of the assisted LSPIV workflow using the proposed tools leads to significant improvements of the discharge measurements with strong reductions of the inter‐participant variability. On the eight videos, the mean interquartile range of the discharge errors is reduced from 17% to 5% and the mean discharge bias is reduced from −9% to 1% with the assisted LSPIV workflow. The remaining inter‐participant variability is mainly due to the user‐defined surface velocity coefficient α.

Automatic filter pruning algorithm for image classification

Automatic filter pruning algorithm for image classification

Pipeline Design for Data Preparation for Social Media Analysis

In a data-driven culture, in which analytics applications are the main resources for supporting decision-making, the use of high-quality datasets is mandatory to minimize errors and risks. For this reason, data analysis tasks need to be preceded by a data preparation pipeline. The design of such a pipeline is not trivial: the data analyst must carefully choose the appropriate operations considering several aspects. This is often performed by adopting a trial-and-error approach that does not always lead to the most effective solution. In addition, extracting information from social media poses specific problems due to the need to consider only posts relevant for the analysis, for its dependence from the context being considered, for its multimedia contents, and for the risk of filtering out informative posts with automatic filters. In this article, we propose a systematic approach to support the design of pipelines that are able to effectively extract a relevant dataset for the goal of the analysis of data from social media. We provide a conceptual model for designing and annotating the data preparation pipeline with quality and performance information, thus providing the data analyst preliminary information on the expected quality of the resulting dataset in a context-aware manner. The generation of metadata related to the processing tasks has been recognized as essential for enabling data sharing and reusability. To this aim, the dataset resulting from the pipeline application is automatically annotated with provenance metadata to get a detailed description of all the activities performed by the pipeline on them. As a case study, we consider the design of a pipeline for creating datasets of images extracted from social media in order to analyze behavioural aspects during COVID-19.

EVALUATION OF THE THORACIC COMPUTED TOMOGRAPHY TECHNIQUE IN THE DIAGNOSIS OF PULMONARY EMBOLISM AT THE LUXEMBOURG MOTHER AND CHILD UNIVERSITY HOSPITAL IN BAMAKO

Introduction: The aim of our study was to evaluate the technique of thoracic computed tomography in the diagnosis of pulmonary embolism at the University Hospital Mère-Enfant le Luxembourg in Bamako. Materials and Methods: This was a descriptive, prospective, cross-sectional study from 1 February to 30 April 2021, a period of 3 months. All patients of any age and sex referred to the radiology department for thoracic computed tomographywho had completed the administrative formalities and agreed to take part in the study, were included in our study. The size of our sample was 46 patients. Data were collected using individual survey forms prepared with the Epi. Info. Version 7.2 tool and filled in on the basis of examination bulletins completed by questioning the patient or accompanying person. Results: At the end of our study in the medical imaging department of the University Hospital Mère-Enfant le Luxembourg in Bamako, the prevalence of pulmonary embolism was 30%. The physical and psychological preparation was 100%, the venous route of the back of the hand with catheter ≤ 20G was the most used 56.52%, the means of restraint were used to 71.7%. Apnoea after deep inspiration of the patient without Valsalva was 84.78%, the automatic 3D reconstruction filter was used in 100% of examinations. The most commonly used PCI was telebris35 (63%) and 45.7% of examinations (n=21) were performed with a volume of 90ml of PCI with an injection rate of 3.5ml/s, i.e. 56.5%. The automatic injector was used in 100% of the examinations, then 81% of the examinations in adults compared with 100% in children (n=4) were performed with a PDL greater than the NRD but neither children nor adults had benefited from any means of radiation protection during the examination. The success rate for a single scan without revision was 52.2%, and the success rate for an examination with good vascular opacification was 95.7%. Conclusion: Thoracic angiography is a key examination in the etiological assessment of pulmonary embolism. It was in high demand during our study period because of the COVID-19 pandemic. In short, thoracic pathologies are becoming increasingly common. The most frequent reason was the onset of pain. However, an early diagnosis is essential for a better prognosis. Angioscan remains the reference examination for vascular pathologies such as pulmonary embolism.

Social Media Content Moderation Challenges for Vulnerable Groups: A Case Study on Tiktok Indonesia

As a popular social media platform, TikTok relies heavily on User-Generated Content (UGC), which is also one of its main attractions. The rapid production of UGC has implications for Tiktok's ability to moderate a large amount of content in a short time. Even though algorithmic technology and automatic filtering are used to assist this process, a moderator team that is responsive and efficient in carrying out moderation actions is needed. This need has implications for the growth of the Commercial Content Moderation (CCM) industry, where Tiktok employs a team of moderators to check and moderate content uploaded by users. The research aims to explore the complexities and unique aspects of Indonesian Tiktok content moderation and its implications for the development of UGC. This research uses an in-depth interview method approach with Tiktok Moderator Team informants as primary data, as well as related secondary uses as supporting data. The research focuses on how moderation practices are carried out by the Indonesian Tiktok Moderator Team in handling content related to vulnerable groups in society such as women, children, and the elderly, who in the last seven years have colored the dynamics of Tiktok in Indonesia. By highlighting the challenges faced by the Moderator Team in finding a balance between giving users freedom of expression and maintaining the security of the Tiktok platform, this research offers recommendations for improving content moderation practices on the platform, especially for the protection of these vulnerable groups.

When Automatic Filtering Comes to the Rescue: Pre-Computing Company Competitor Pairs in Owler

Competitor data constitutes information significantly valuable for many business applications. Meltwater provides users with access to a large Company Information System (CIS), Owler, which contains competitor pairs and other useful information about companies. Meltwater has been seeking a practical solution to discover more competitor pairs in Owler. The first attempt, a fully-manual workflow (called MW_Manual) for finding more competitor pairs in Owler consisted of two manual steps: a filtering step that excludes obvious non-competitor company pairs, and a further inspection process that inspects each left company pair after the filtering step. MW_Manual was cost prohibitive because the results of the filtering step contained too many non-competitor pairs. Inspecting such non-competitor pairs caused an overhead to the overall workload. To reduce the manual workload, especially the required human effort in the manual inspection process, Meltwater has transformed MW_Manual into a semi-automatic workflow (called MW_CPFilter) by replacing the manual filtering with an automatic yet more precise process that adopts a system called CPFilter. This paper presents CPFilter, a system used in the filtering process of MW_CPFilter. CPFilter automatically pre-computes likely competitor pairs from existing competitor pairs in Owler. CPFilter combines (i) the generation of new competitor candidate pairs by inference from existing competitors and other company-specific knowledge, with (ii) the validation of each candidate competitor pair of two companies by checking whether or not empirical evidence that indicates the competitor relationships of these two companies can be found. CPFilter has three key advantages compared with the manual filtering process and previous works: (i) it resulted in a high workload reduction rate of 0.81, (ii) it is domain-independent so that it can be applied to different sectors in Owler, and (iii) its results are explainable so that humans can easily understand its results.

Agile Methodologies Applied to the Development of Internet of Things (IoT)-Based Systems: A Review.

Throughout the evolution of software systems, empirical methodologies have been used in their development process, even in the Internet of Things (IoT) paradigm, to develop IoT-based systems (IoTS). In this paper, we review the fundamentals included in the manifesto for agile software development, especially in the Scrum methodology, to determine its use and role in IoTS development. Initially, 4303 documents were retrieved, a number that was reduced to 186 after applying automatic filters and by the relevance of their titles. After analysing their contents, only 60 documents were considered. Of these, 38 documents present the development of an IoTS using some methodology, 8 present methodologies focused on the construction of IoTS software, and 14 present methodologies close to the systems life cycle (SLC). Finally, only one methodology can be considered SLC-compliant. Out of 38 papers presenting the development of some IoTS following a methodology for traditional information systems (ISs), 42.1% have used Scrum as the only methodology, while 10.5% have used Scrum combined with other methodologies, such as eXtreme Programming (XP), Kanban and Rapid Prototyping. In the analysis presented herein, the existing methodologies for developing IoTSs have been grouped according to the different approaches on which they are based, such as agile, modelling, and service oriented. This study also analyses whether the different proposals consider the standard stages of the development process or not: planning and requirements gathering, solution analysis, solution design, solution coding and unit testing (construction), integration and testing (implementation), and operation and maintenance. In addition, we include a review of the automated frameworks, platforms, and tools used in the methodologies analysed to improve the development of IoTSs and the design of their underlying architectures. To conclude, the main contribution of this work is a review for IoTS researchers and developers regarding existing methodologies, frameworks, platforms, tools, and guidelines for the development of IoTSs, with a deep analysis framed within international standards dictated for this purpose.

Early detection of tuberculosis using hybrid feature descriptors and deep learning network.

To detect tuberculosis (TB) at an early stage by analyzing chest X-ray images using a deep neural network, and to evaluate the efficacy of proposed model by comparing it with existing studies. For the study, an open-source X-ray images were used. Dataset consisted of two types of images, i.e., standard and tuberculosis. Total number of images in the dataset was 4,200, among which, 3,500 were normal chest X-rays, and the remaining 700 X-ray images were of tuberculosis patients. The study proposed and simulated a deep learning prediction model for early TB diagnosis by combining deep features with hand-engineered features. Gabor filter and Canny edge detection method were applied to enhance the performance and reduce computation cost. The proposed model simulated two scenarios: without filter and edge detection techniques and only a pre-trained model with automatic feature extraction, and filter and edge detection techniques. The results achieved from both the models were 95.7% and 97.9%, respectively. The proposed study can assist in the detection if a radiologist is not available. Also, the model was tested with real-time images to examine the efficacy, and was better than other available models.

Iterative Reconstruction: State-of-the-Art and Future Perspectives.

Image reconstruction processing in computed tomography (CT) has evolved tremendously since its creation, succeeding at optimizing radiation dose while maintaining adequate image quality. Computed tomography vendors have developed and implemented various technical advances, such as automatic noise reduction filters, automatic exposure control, and refined imaging reconstruction algorithms.Focusing on imaging reconstruction, filtered back-projection has represented the standard reconstruction algorithm for over 3 decades, obtaining adequate image quality at standard radiation dose exposures. To overcome filtered back-projection reconstruction flaws in low-dose CT data sets, advanced iterative reconstruction algorithms consisting of either backward projection or both backward and forward projections have been developed, with the goal to enable low-dose CT acquisitions with high image quality. Iterative reconstruction techniques play a key role in routine workflow implementation (eg, screening protocols, vascular and pediatric applications), in quantitative CT imaging applications, and in dose exposure limitation in oncologic patients.Therefore, this review aims to provide an overview of the technical principles and the main clinical application of iterative reconstruction algorithms, focusing on the strengths and weaknesses, in addition to integrating future perspectives in the new era of artificial intelligence.

Validity of a Smartphone Application in Calculating Measures of Heart Rate Variability.

The purpose of the current study was to determine the concurrent validity of the Elite HRV smartphone application when calculating heart rate variability (HRV) metrics in reference to an independent software criterion. A total of 5 minutes of R−R interval and natural log of root mean square of the successive differences (lnRMSSD) resting HRV data were simultaneously collected using two Polar H10 heart rate monitors (HRMs) in both the seated and supine positions from 22 participants (14 males, 8 females). One H10 HRM was paired with a Polar V800 watch and one with the Elite HRV application. When no artifact correction was applied, significant, but small, differences in the lnRMSSD data were observed between the software in the seated position (p = 0.022), and trivial and nonstatistically significant differences were observed in the supine position (p = 0.087). However, significant differences (p > 0.05) in the lnRMSSD data were no longer identifiable in either the seated or the supine positions when applying Very Low, Low, or Automatic artifact-correction filters. Additionally, excellent agreements (ICC3,1 = 0.938 − 0.998) and very strong to near-perfect (r = 0.889 − 0.997) relationships were observed throughout all correction levels. The Elite HRV smartphone application is a valid tool for calculating resting lnRMSSD HRV metrics.

Automatic Filtering and Classification of Low-Density Airborne Laser Scanner Clouds in Shrubland Environments

The monitoring of shrublands plays a fundamental role, from an ecological and climatic point of view, in biodiversity conservation, carbon stock estimates, and climate-change impact assessments. Laser scanning systems have proven to have a high capability in mapping non-herbaceous vegetation by classifying high-density point clouds. On the other hand, the classification of low-density airborne laser scanner (ALS) clouds is largely affected by confusion with rock spikes and boulders having similar heights and shapes. To identify rocks and improve the accuracy of vegetation classes, we implemented an effective and time-saving procedure based on the integration of geometric features with laser intensity segmented by K-means clustering (GIK procedure). The classification accuracy was evaluated, taking into account the data unevenness (small size of rock class vs. vegetation and terrain classes) by estimating the Balanced Accuracy (BA range 89.15–90.37); a comparison with a standard geometry-based procedure showed an increase in accuracy of about 27%. The classical overall accuracy is generally very high for all the classifications: the average is 92.7 for geometry-based and 94.9 for GIK. At class level, the precision (user’s accuracy) for vegetation classes is very high (on average, 92.6% for shrubs and 99% for bushes) with a relative increase for shrubs up to 20% (>10% when rocks occupy more than 8% of the scene). Less pronounced differences were found for bushes (maximum 4.13%). The precision of rock class is quite acceptable (about 64%), compared to the complete absence of detection of the geometric procedure. We also evaluated how point cloud density affects the proposed procedure and found that the increase in shrub precision is also preserved for ALS clouds with very low point density (<1.5 pts/m2). The easiness of the approach also makes it implementable in an operative context for a non-full expert in LiDAR data classification, and it is suitable for the great wealth of large-scale acquisitions carried out in the past by using monowavelength NIR laser scanners with a small footprint configuration.