Feature Detection Research Articles

Use of artificial intelligence for recognition of biomarkers in intermediate age-related macular degeneration

Advances in imaging and artificial intelligence (AI) have revolutionized the detection, quantification and monitoring for the clinical assessment of intermediate age-related macular degeneration (iAMD). The iAMD incorporates a broad spectrum of manifestations, which range from individual small drusen, hyperpigmentation, hypopigmentation up to early stages of geographical atrophy. Current high-resolution imaging technologies enable an accurate detection and description of anatomical features, such as drusen volumes, hyperreflexive foci and photoreceptor degeneration, which are risk factors that are decisive for prediction of the course of the disease; however, the manual annotation of these features in complex optical coherence tomography (OCT) scans is impractical for the routine clinical practice and research. In this context AI provides a solution by fully automatic segmentation and therefore delivers exact, reproducible and quantitative analyses of AMD-related biomarkers. Furthermore, the application of AI in iAMD facilitates the risk assessment and the development of structural endpoints for new forms of treatment. For example, the quantitative analysis of drusen volume and hyperreflective foci with AI algorithms has shown a correlation with the progression of the disease. These technological advances therefore improve not only the diagnostic precision but also support future targeted treatment strategies and contribute to the prioritized target of personalized medicine in the diagnostics and treatment of AMD.

Detection of subsurface archaeological features using the GPR method with a 250 MHz antenna in Borsippa site, Babylon, Iraq

The GPR method was used to determine the depth and extent of the subsurface archaeological features at the archaeological site of Borsippa. The extensions of the archaeological walls were identified through the creation of three-dimensional maps. The LMX200 device conducted the survey using a 250 MHZ antenna. sixty parallel profiles were collected and processed by GPR Slice software. Several reflections of possible archaeological walls buried at different depths and extensions of 2, 4, 6, 11, and 20 m were identified. The 250 MHz antenna gave clear reflections of the archaeological walls, with a penetration depth of up to 4 meters. The time slice map identified a buried archaeological structure at 1–1.1 m deep, consisting of many rooms with thick walls up to 2-3 m.

Robotic Grasping of Unknown Objects Based on Deep Learning-Based Feature Detection.

In recent years, the integration of deep learning into robotic grasping algorithms has led to significant advancements in this field. However, one of the challenges faced by many existing deep learning-based grasping algorithms is their reliance on extensive training data, which makes them less effective when encountering unknown objects not present in the training dataset. This paper presents a simple and effective grasping algorithm that addresses this challenge through the utilization of a deep learning-based object detector, focusing on oriented detection of key features shared among most objects, namely straight edges and corners. By integrating these features with information obtained through image segmentation, the proposed algorithm can logically deduce a grasping pose without being limited by the size of the training dataset. Experimental results on actual robotic grasping of unknown objects over 400 trials show that the proposed method can achieve a higher grasp success rate of 98.25% compared to existing methods.

A Deep Learning Lidar Denoising Approach for Improving Atmospheric Feature Detection

Space-based atmospheric backscatter lidars provide critical information about the vertical distribution of clouds and aerosols, thereby improving our understanding of the climate system. They are additionally useful for detecting hazards to aviation and human health, such as volcanic plumes and man-made pollution events. The Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP, 2006–2023), Cloud-Aerosol Transport System (CATS, 2015–2017), and Advanced Topographic Laser Altimeter System (ATLAS 2018–present) are three such lidars that operated within the past 20 years. The signal-to-noise ratio (SNR) for these lidars is significantly lower in daytime data compared with nighttime data due to the solar background signal increasing the detector response noise. Averaging horizontally across profiles has been the standard way to increase SNR, but this comes at the expense of resolution. Modern, deep learning-based denoising algorithms can be applied to improve the SNR without coarsening resolution. This paper describes how one such model architecture, Dense Dense U-Net (DDUNet), was trained to denoise CATS 1064 nm raw signal data (photon counts) using artificially noised nighttime data. Simulated CATS daytime 1064 nm data were then created to assess the model’s performance. The denoised simulated data increased the daytime SNR by a factor of 2.5 (on average) and decreased minimum detectable backscatter (MDB) to ~7.3×10−4 km−1sr−1, which is lower than the CALIOP 1064 nm night MDB value of 8.6×10−4 km−1sr−1. Layer detection was performed on simulated 2 km horizontal resolution denoised and 60 km averaged data. Despite the finer resolution input, the denoised layers had more true positives, fewer false positives, and an overall Jaccard Index of 0.54 versus 0.44 when compared to the layers detected on averaged data. Layer detection was also performed on a full month of denoised daytime CATS data (Aug. 2015) to detect layers for comparison with CATS standard Level 2 (L2) product layers. The detection on the denoised data yielded 2.33 times more, higher-quality bins within detected layers at 2.7–33 times finer resolution than the CATS L2 products.

Feature Detection of Non-Cooperative and Rotating Space Objects through Bayesian Optimization.

In this paper, we propose a Bayesian Optimization (BO)-based strategy using the Gaussian Process (GP) for feature detection of a known but non-cooperative space object by a chaser with a monocular camera and a single-beam LIDAR in a close-proximity operation. Specifically, the objective of the proposed Space Object Chaser-Resident Assessment Feature Tracking (SOCRAFT) algorithm is to determine the camera directional angles so that the maximum number of features within the camera range is detected while the chaser moves in a predefined orbit around the target. For the chaser-object spatial incentive, rewards are assigned to the chaser states from a combined model with two components: feature detection score and sinusoidal reward. To calculate the sinusoidal reward, estimated feature locations are required, which are predicted by Gaussian Process models. Another Gaussian Process model provides the reward distribution, which is then used by the Bayesian Optimization to determine the camera directional angles. Simulations are conducted in both 2D and 3D domains. The results demonstrate that SOCRAFT can generally detect the maximum number of features within the limited camera range and field of view.

DMWMNet: A novel dual-branch multi-level convolutional network for high-performance mixed-type wafer map defect detection in semiconductor manufacturing

Wafer map defect detection plays an important role in semiconductor manufacturing by identifying root causes and accelerating process adjustments to ensure product quality and reduce unnecessary expenditures. However, existing methods have some limitations, such as low accuracy in mixed-type defect detection and poor recognition of similar defects and weak features. In this article, a novel dual-branch multi-level convolutional network (DMWMNet) is proposed for high-performance mixed-type wafer map defect detection. By fully considering the interrelationships between basic defects, defect number, and defect type, the network is designed to include two efficient parallel Branches and a Fusion classifier. Detecting defect types using basic defect discrimination and defect number detection is helpful for ameliorating problems with high complexity and low accuracy caused by multiple defect categories and feature overlaps. Furthermore, a composite loss function based on focal loss is employed to improve the network’s capacity to recognize weak features and similar defects. Experimental results on the MixedWM38 dataset show that DMWMNet has favorable mixed-type defect detection performance compared to other methods, with accuracy, precision, recall, F1 score, and MCC of 98.99%, 98.94%, 99.03%, 98.98%, and 98.97%, respectively.

Spectral feature informed variational model and its applications to machinery fault diagnosis

Variational mode extraction (VME), a novel signal decomposition method based on a frequency-domain filter in essence, has recently become a potential tool in fault diagnosis. However, the original VME algorithm is not provided with full self-adaptation, and its performance in the extraction of fault features is subject to predefining the initial parameters, including initial center frequency (ICF) and balance parameter. To address these issues, a spectral feature informed variational model (SFIVM) algorithm is constructed to overcome the defects of parameters setting and efficiently realize the fault diagnosis without prior knowledge. Specifically, a spectral feature detector inspired by the convergence property of ICF is first developed to reveal the spectral features, including the detected center frequencies and boundary frequencies. Then, a balance parameter estimation formula is designed to adaptively determine the target balance parameter by taking advantage of the above spectral features. Finally, a highly efficient decomposition model is proposed to extract the fault-related mode from the vibration signal, where iterative optimization is unnecessary. The effectiveness of the proposed SFIVM method is verified by one simulated and two experimental cases. Moreover, its superiority and high efficiency are demonstrated by comparing it with some advanced and classical fault diagnosis methods.

How the Degree of Anthropomorphism of Human-like Robots Affects Users' Perceptual and Emotional Processing: Evidence from an EEG Study.

Anthropomorphized robots are increasingly integrated into human social life, playing vital roles across various fields. This study aimed to elucidate the neural dynamics underlying users' perceptual and emotional responses to robots with varying levels of anthropomorphism. We investigated event-related potentials (ERPs) and event-related spectral perturbations (ERSPs) elicited while participants viewed, perceived, and rated the affection of robots with low (L-AR), medium (M-AR), and high (H-AR) levels of anthropomorphism. EEG data were recorded from 42 participants. Results revealed that H-AR induced a more negative N1 and increased frontal theta power, but decreased P2 in early time windows. Conversely, M-AR and L-AR elicited larger P2 compared to H-AR. In later time windows, M-AR generated greater late positive potential (LPP) and enhanced parietal-occipital theta oscillations than H-AR and L-AR. These findings suggest distinct neural processing phases: early feature detection and selective attention allocation, followed by later affective appraisal. Early detection of facial form and animacy, with P2 reflecting higher-order visual processing, appeared to correlate with anthropomorphism levels. This research advances the understanding of emotional processing in anthropomorphic robot design and provides valuable insights for robot designers and manufacturers regarding emotional and feature design, evaluation, and promotion of anthropomorphic robots.

RCRFNet: Enhancing Object Detection with Self-Supervised Radar-Camera Fusion and Open-Set Recognition.

Robust object detection in complex environments, poor visual conditions, and open scenarios presents significant technical challenges in autonomous driving. These challenges necessitate the development of advanced fusion methods for millimeter-wave (mmWave) radar point cloud data and visual images. To address these issues, this paper proposes a radar-camera robust fusion network (RCRFNet), which leverages self-supervised learning and open-set recognition to effectively utilise the complementary information from both sensors. Specifically, the network uses matched radar-camera data through a frustum association approach to generate self-supervised signals, enhancing network training. The integration of global and local depth consistencies between radar point clouds and visual images, along with image features, helps construct object class confidence levels for detecting unknown targets. Additionally, these techniques are combined with a multi-layer feature extraction backbone and a multimodal feature detection head to achieve robust object detection. Experiments on the nuScenes public dataset demonstrate that RCRFNet outperforms state-of-the-art (SOTA) methods, particularly in conditions of low visual visibility and when detecting unknown class objects.

Close-range imaging for green roofs: Feature detection, band matching, and image registration for mixed plant communities

Green roofs offer ecological benefits but harsh rooftop environment stressors like heat, water scarcity, and wind can limit optimal plant growth. Monitoring plant health is crucial for optimizing green roof performance and remote sensing provides a non-destructive approach, yet challenges persist in tight, urban settings. This allows for close-range ground monitoring as a viable solution in these emerging environments. Recent studies explored automatic image registration techniques, showing promise in homogeneous settings, but faced uncertainties in mixed species systems or early plant growth stages. Feature detection techniques such as Speeded up Robust Features (SURF) using fixed and moving image spaces have been proposed for alignment. This study aimed to assess feature detection, matching, and image registration techniques in mixed species plant communities on green roofs throughout the growing season. This study developed a sophisticated monitoring system using close-range multispectral sensors for aligning bands in mixed species plant communities, contributing to enhanced green roof management and sustainability. The results of the study showed no significant differences in band alignment accuracy between growth stages for mixed species bush bean/sedum and single species sedum modules, but significant differences were found in single species bush bean systems. Additionally, mixed species modules showed better accuracy compared to single species bush bean modules, indicating that heterogenous systems provide better alignment accuracy due to more diverse feature extractions.

A Robust AR-DSNet Tracking Registration Method in Complex Scenarios

A robust AR-DSNet (Augmented Reality method based on DSST and SiamFC networks) tracking registration method in complex scenarios is proposed to improve the ability of AR (Augmented Reality) tracking registration to distinguish target foreground and semantic interference background, and to address the issue of registration failure caused by similar target drift when obtaining scale information based on predicted target positions. Firstly, the pre-trained network in SiamFC (Siamese Fully-Convolutional) is utilized to obtain the response map of a larger search area and set a threshold to filter out the initial possible positions of the target; Then, combining the advantage of the DSST (Discriminative Scale Space Tracking) filter tracker to update the template online, a new scale filter is trained after collecting multi-scale images at the initial possible position of target to reason the target scale change. And linear interpolation is used to update the correlation coefficient to determine the final position of target tracking based on the difference between two frames. Finally, ORB (Oriented FAST and Rotated BRIEF) feature detection and matching are performed on the accurate target position image, and the registration matrix is calculated through matching relationships to overlay the virtual model onto the real scene, achieving enhancement of the real world. Simulation experiments show that in complex scenarios such as similar interference, target occlusion, and local deformation, the proposed AR-DSNet method can complete the registration of the target in AR 3D tracking, ensuring real-time performance while improving the robustness of the AR tracking registration algorithm.

Broadscale deep learning model for archaeological feature detection across the Maya area

Many Maya archaeological areas are not comprehensively or systematically mapped because ruins, often hidden under tropical forest canopy in rugged terrain, can take decades to locate, identify, and map. Recent years have seen an explosion of lidar data collection, and machine learning provides a way to exploit these lidar data, making feature analyses more efficient and consistently executed. At present, there are a limited number of small, area-specific models that exist for the Maya area, the largest of which covers 230 km2. Here we present the foundation for a broadscale, multi-area-based convolutional neural network (CNN) object detection model that uses airborne laser scanning data, or lidar, for archaeological feature detection across 615 km2 of the Maya area, as well as preliminary results from an additional 885 km2 test area. This sets the path for a model that will enable researchers to map archaeological areas across the entire Maya Lowland area in weeks or months instead of decades. Notably, we find that a model trained on multiple areas with significantly different topographies produces better results for all areas as compared to a model trained on a single area. The broadscale model here presented produced an F1 score of 0.80. Results also include many potential new structure detections, including detections on lidar at an archaeological area that has not yet been comprehensively ground-surveyed and is located in an entirely different country in the Maya Lowlands from where the model was trained on. This model represents an attempt at a broadscale machine learning approach for archaeological feature mapping in the Maya area and demonstrates how big data can be integrated into traditional archaeological workflows. Lidar has already shown much greater ancient Maya infrastructure throughout the Maya world and elsewhere in the tropics, and this study using machine learning with lidar is showing even greater Maya infrastructure through vast areas of the Maya tropical forest.

Classification of Time–Frequency Maps of Guided Waves Using Foreground Extraction

Guided waves propagating in mechanical structures have proved to be an essential technique for applications, such as structural health monitoring. However, it is a well-known problem that when using non-stationary guided wave signals, dispersion, and high-order vibrational modes are excited, it becomes cumbersome to detect and identify relevant information. A typical method for the characterization of these non-stationary signals is based on time–frequency (TF) mapping techniques. This method produces 2D images, allowing the study of specific vibration modes and their evolution over time. However, this approach has low resolution, increases the size of the data, and introduces redundant information, making it difficult to extract relevant features for their accurate identification and classification. This paper presents a method for identifying discontinuities by analyzing the data in the TF maps of Lamb wave signals. Singular Value Decomposition (SVD) for low-rank optimization and then perform foreground feature extraction on the maps were proposed. These foreground features are then analyzed using Principal Component Analysis (PCA). Unlike traditional PCA, which operates on vectorized images, our approach focuses on the correlation between coordinates within the maps. This modification enhances feature detection and enables the classification of discontinuities within the maps. To evaluate unsupervised clustering of the dimensionally reduced data obtained from PCA, we experimentally tested our method using broadband Lamb waves with various vibrational modes interacting with different types of discontinuity patterns in a thin aluminum plate. A Support Vector Machine (SVM) classifier was then implemented for classification. The results of the experimental data yielded good classification effectiveness within reasonably low computational time despite the large matrixes of the TF maps used.

A stochastic approach for parameter optimization of feature detection algorithms for non-target screening in mass spectrometry.

Feature detection plays a crucial role in non-target screening (NTS), requiring careful selection of algorithm parameters to minimize false positive (FP) features. In this study, a stochastic approach was employed to optimize the parameter settings of feature detection algorithms used in processing high-resolution mass spectrometry data. This approach was demonstrated using four open-source algorithms (OpenMS, SAFD, XCMS, and KPIC2) within the patRoon software platform for processing extracts from drinking water samples spiked with 46 per- and polyfluoroalkyl substances (PFAS). The designed method is based on a stochastic strategy involving random sampling from variable space and the use of Pearson correlation to assess the impact of each parameter on the number of detected suspect analytes. Using our approach, the optimized parameters led to improvement in the algorithm performance by increasing suspect hits in case of SAFD and XCMS, and reducing the total number of detected features (i.e., minimizing FP) for OpenMS. These improvements were further validated on three different drinking water samples as test dataset. The optimized parameters resulted in a lower false discovery rate (FDR%) compared to the default parameters, effectively increasing the detection of true positive features. This work also highlights the necessity of algorithm parameter optimization prior to starting the NTS to reduce the complexity of such datasets.

Tobac v1.5: introducing fast 3D tracking, splits and mergers, and other enhancements for identifying and analysing meteorological phenomena

Abstract. There is a continuously increasing need for reliable feature detection and tracking tools based on objective analysis principles for use with meteorological data. Many tools have been developed over the previous 2 decades that attempt to address this need but most have limitations on the type of data they can be used with, feature computational and/or memory expenses that make them unwieldy with larger datasets, or require some form of data reduction prior to use that limits the tool's utility. The Tracking and Object-Based Analysis of Clouds (tobac) Python package is a modular, open-source tool that improves on the overall generality and utility of past tools. A number of scientific improvements (three spatial dimensions, splits and mergers of features, an internal spectral filtering tool) and procedural enhancements (increased computational efficiency, internal regridding of data, and treatments for periodic boundary conditions) have been included in tobac as a part of the tobac v1.5 update. These improvements have made tobac one of the most robust, powerful, and flexible identification and tracking tools in our field to date and expand its potential use in other fields. Future plans for tobac v2 are also discussed.

Biological Basis and Computer Vision Applications of Image Phase Congruency: A Comprehensive Survey.

The concept of Image Phase Congruency (IPC) is deeply rooted in the way the human visual system interprets and processes spatial frequency information. It plays an important role in visual perception, influencing our capacity to identify objects, recognize textures, and decipher spatial relationships in our environments. IPC is robust to changes in lighting, contrast, and other variables that might modify the amplitude of light waves yet leave their relative phase unchanged. This characteristic is vital for perceptual tasks as it ensures the consistent detection of features regardless of fluctuations in illumination or other environmental factors. It can also impact cognitive and emotional responses; cohesive phase information across elements fosters a perception of unity or harmony, while inconsistencies can engender a sense of discord or tension. In this survey, we begin by examining the evidence from biological vision studies suggesting that IPC is employed by the human perceptual system. We proceed to outline the typical mathematical representation and different computational approaches to IPC. We then summarize the extensive applications of IPC in computer vision, including denoise, image quality assessment, feature detection and description, image segmentation, image registration, image fusion, and object detection, among other uses, and illustrate its advantages with a number of examples. Finally, we discuss the current challenges associated with the practical applications of IPC and potential avenues for enhancement.

Face Detection and Recognition for Criminal Identification System

This project showcases a powerful Face Detection and Recognition system intended to improve law enforcement skills by speeding up the identification of suspects in criminal cases. The system uses scalable databases and real-time computation, however instead of using traditional web sources for face detection. Accuracy and dependability are ensured by using the LBPH algorithm for face recognition and the Haar-cascade classifier for feature detection. The smtplib library allows the system to send a Gmail notification when it finds a matching face. The system that has been built contributes to public safety initiatives by improving law enforcement operations and demonstrating versatility for different face detection circumstances. The outcomes show how successfully the system can detect faces in real-time circumstances. The system achieves great performance by creating dynamic datasets and integrating algorithms. Integrating Gmail improves law enforcement response by facilitating prompt contact of identified suspects. This study highlights the role that surveillance technology plays in public safety and crime prevention, and it provides law enforcement authorities with useful solutions. Furthermore, the uses of this facial recognition and detection technology go beyond police enforcement. It can expedite attendance tracking and improve security in access control systems. Because of its adaptability, it can be used in a variety of industries and is expected to have a significant impact.

Method for determining the types of encryption of linear block codes

Problem statement. All known methods for determining the parameters of the encrypted code are based on known types of channel encoding and encryption, but the scrambling code data received by the uncoordinated communication parties is unknown, therefore it is necessary to analyze whether the outgoing signal is encoded, and which encryption method is adopted, contributing to further parameter determination. Goal. Development of a methodology for determining the types of encryption of linear block codes by combining correlation features and convolutional neural networks. Results. The simulation results show that compared with the traditional feature detection method, this method, based on a convolutional neural network and long short-term memory, has a better recognition speed at BER and can recognize the type of encryption of linear block code in BER and noise conditions. Practical significance. The identification of the encryption code parameters is used to subsequently determine the parameters of the encoder and has significant practical engineering value. The work was carried out with the financial support of the Ministry of Science and Higher Education of the Russian Federation within the framework of the state assignment "youth laboratory" No. FZGM-2024-0003.

3D point cloud reconstruction using panoramic images

Panorama photogrammetry, the process of analyzing panoramic images, has gained popularity in close-range photogrammetry for 3D reconstruction over the past decade. Initially, researchers utilized cylindrical or spherical panoramic images created through specialized cameras or conventional ones with rectilinear lenses. However, these methods were hindered by the high cost of panorama equipment and the need for manual reconstruction. Consequently, there's a growing demand for automated algorithms capable of reconstructing 3D point clouds from stitched panorama images. This study aims to provide a cost-effective solution for automatic 3D point cloud reconstruction from panoramas. The study is divided into two parts; it first outlines an image acquisition strategy for capturing overlapping perspective images to facilitate accurate panorama generation. Subsequently, it introduces an automated algorithm for 3D point cloud reconstruction from panorama images. The process utilizes the KAZE feature detector for feature detection and introduces a novel feature matching approach for matching panorama images. Accuracy assessment of the reconstructed 3D point clouds was done using three methods: Line Segment Based approach, yielding RMSE errors of 34.2mm and 39mm for dataset-1 and dataset-2 respectively, No-Reference 3D Point Cloud Quality Assessment, resulting in quality scores of 8.5939 and 7.4535 for dataset-1 and dataset-2 respectively, and M3C2 distance method computed value of 0.091059 and 0.165179 respectively, indicating high quality of the generated point clouds.

Autonomous and reliable fingerprint map maintenance for indoor positioning system

In WiFi-based crowdsourced fingerprinting indoor positioning systems (IPS), the signal space of WiFi APs changes over time. To autonomously maintain a highly reliable localization accuracy, the fingerprint database needs regular maintenance and updates. However, this process is labor-intensive and time-consuming due to the appearance of faulty access points (APs) in crowdsourced samples received from target clients, which are mainly used for autonomous maintenance and updates. Additionally, if some APs are relocated, removed, or replaced in the indoor environment after fingerprint map construction, it can severely impact IPS accuracy. In this paper, we present Dynamic Online Fingerprint Learning with Altered APs status-aware and updating (DOFLAAU) for IPS. Our systematic methodology involves several key steps. First, we employ the Modified Group Method of Data Handling (GMDH) neural network regression for autonomous AP status monitoring and alteration detection. Next, we perform autonomous APs subset sample selection using a modified Thomson model, followed by outlier feature detection and removal based on the weighted sample density rate. Thirdly, we propose a dynamic online fingerprint map updating model using the cosine similarity distance metric. Lastly, we adopted an effective signal distance-based weighted k-nearest neighbor algorithm to improve IPS accuracy. The experiment results validate the efficiency and effectiveness of our proposed algorithms in terms of better localization accuracy compared to peer algorithms.