Camera System Research Articles

Camera System to Monitor Residential Societies Vehicle Activities

Camera System to Monitor Residential Societies Vehicle Activities

An optimized lens design using optical imaging for surveillance camera systems with minimal aberration

An optimized lens design using optical imaging for surveillance camera systems with minimal aberration

Monitoring Vehicle Activity in Residential Societies

An increase in vehicles within residential societies has resulted in issues, including congestions during parking, unpermitted vehicle access, and raised security concerns. For such issues, this paper has looked into developing a low-cost mobile application camera system intended to track the vehicle activity in such communities. Low in terms of cost, efficient with regards to vehicle management as well as security, and has the ability to utilize free-of-charge mobile technologies in tandem with strategically place camera systems around the community, which can track a vehicle's movement and provide on-the-spot recording and indication of unauthorized entry with respective alerts to residents and relevant security personnel. The mobile application is also a user-friendly interface by which residents can monitor the real-time footage, can receive alerts, and easily manage access to parking areas. Moreover, the system promotes community engagement by allowing participation in the monitoring process. It enhances vigilance generally. This solution is not only affordable but also scalable, thus accessible to residential societies of a wide variety. This would enable the system to focus more on cost-effectiveness and user-friendliness while, at the same time, improving security, vehicle management, and contributing to an organized, safer living space for its residents.

Evaluating the Reliability and Consistency of Treadmill Gait Analysis Using an RGB-D Camera: Effects of Assistance and No Assistance.

This study aimed to assess the intraday reliability of markerless gait analysis using an RGB-D camera versus a traditional three-dimensional motion analysis (3DMA) system with and without a simulated walking assistant. Gait assessments were conducted on 20 healthy adults walking on a treadmill with a focus on spatiotemporal parameters gathered using the RGB-D camera and 3DMA system. The intraday reliability of the RGB-D camera was evaluated using intraclass correlation coefficients (ICC 1, 1), while its consistency with the 3DMA system was determined using ICC (2, 1). The results demonstrated that the RGB-D camera provided high intraday reliability and showed strong consistency with 3DMA measurements regardless of the presence of an assistant. The Bland-Atman analysis indicated no significant systematic bias, with the minimum detectable change remaining within acceptable clinical ranges. These findings highlight the potential of the RGB-D camera for reliable markerless gait analysis in clinical environments in which walking assistance may be needed, thereby expanding its applicability in patients with various impairment degrees. Future research should validate these results in patient populations and explore their utility for measuring kinematic parameters.

Event-Based Visual Simultaneous Localization and Mapping (EVSLAM) Techniques: State of the Art and Future Directions

Recent advances in event-based cameras have led to significant developments in robotics, particularly in visual simultaneous localization and mapping (VSLAM) applications. This technique enables real-time camera motion estimation and simultaneous environment mapping using visual sensors on mobile platforms. Event cameras offer several distinct advantages over frame-based cameras, including a high dynamic range, high temporal resolution, low power consumption, and low latency. These attributes make event cameras highly suitable for addressing performance issues in challenging scenarios such as high-speed motion and environments with high-range illumination. This review paper delves into event-based VSLAM (EVSLAM) algorithms, leveraging the advantages inherent in event streams for localization and mapping endeavors. The exposition commences by explaining the operational principles of event cameras, providing insights into the diverse event representations applied in event data preprocessing. A crucial facet of this survey is the systematic categorization of EVSLAM research into three key parts: event preprocessing, event tracking, and sensor fusion algorithms in EVSLAM. Each category undergoes meticulous examination, offering practical insights and guidance for comprehending each approach. Moreover, we thoroughly assess state-of-the-art (SOTA) methods, emphasizing conducting the evaluation on a specific dataset for enhanced comparability. This evaluation sheds light on current challenges and outlines promising avenues for future research, emphasizing the persisting obstacles and potential advancements in this dynamically evolving domain.

Advancements in Autonomous Driving Systems: Enhancing Environmental Perception with Lidar Integration and Innovative Algorithms

This paper discusses the intricacies of autonomous driving systems, focusing on the four core modules: perception, sensing, decision-making, and control. The sensing module integrates various sensors like LiDAR, cameras, millimeter-wave radars, IMU, and GPS to gather essential data. The paper emphasizes the role of LiDAR in 3D object detection and point cloud segmentation, which are crucial for environmental perception. It also addresses the challenge of fusing data from multiple sensors into a unified coordinate system for effective decision-making. This research presents several significant advances in environmental perception. First there is a single-stage 3D object detection algorithm using sparse convolutional neural networks for improved speed and accuracy. Moreover there is a two-stage 3D object detection method focuses on detecting difficult targets through target center prediction. This method employs a novel neural network architecture, AOAP-SCNN, for feature extraction. Furthermore there is a point cloud instance segmentation algorithm (a method that separates and identifies individual objects in 3D space) utilizes a front view approach and a DLA network with Tiny-PointNet for feature extraction. Finally there is also an automatic calibration algorithm for LiDAR and camera systems based on segmentation, designed for real-time operation on embedded platforms, which corrects the extrinsic matrix and improves the self-calibration system's practical value. Building upon these technical innovations, the paper provides a comprehensive analysis of intelligent vehicle technology including the history of development, the importance of environmental perception, and the application and limitations of LiDAR in intelligent vehicles.

Embedded System-Based Image Processing Methods for Detection of Forensic Events in CCTV Videos

Nowadays, there is no place where security cameras (CCTV) are not used. Security cameras play a huge role in solving criminal cases. However, a lot of time is spent examining these camera recordings. This situation causes the incidents to not be resolved and causes delays. This study, it is aimed to use machine learning to increase the size of security camera recordings with efficient algorithms that can work on devices with low processing power such as embedded systems. Within the scope of the study, an experimental environment was created by installing a security camera system. Fast and effective video reduction algorithms have been developed on videos collected in different scenarios. New approaches called hopscotch and lens algorithms have been presented for video reduction. These approaches are aimed to obtain rapid results by applying them to security camera videos. It is thought that the developed video reduction approaches will lead to the creation of applicable prototypes on embedded cards such as Raspberry Pi. PSNR (Peak Signal to Noise Ratio) metric was used to compare the images. Real-time results were obtained with our approaches applied to images.

Biomechanical Behaviors of Molars Restored with Endocrowns Composed of Different Materials.

To assess the biomechanical behaviors of endodontically treated molars (ETMs) restored with endocrowns composed of different materials, forty mandibular molars were assigned to five groups (n = 8 each). Untreated molars constituted the control group (group C); the rest of the teeth that underwent root canal therapy were restored with endocrowns composed of polycrystalline ceramics (ST zirconia®, UPCERA) in group ZR, lithium disilicate glass ceramics (UP.CAD®, UPCERA) in group LD, resin-based nanoceramics (Hyramic®, UPCERA) in group NC, and feldspathic ceramics (CEREC Blocs®, Sirona) in group FC. All teeth were axially loaded until fracture. The process was recorded using a high-speed camera system, and fractographic analysis was conducted. The results showed that fracture loads did not significantly differ among groups C, LD, and NC; the loads were significantly lower than the load in group ZR but higher than the load in group FC. The mean time from the initial crack to complete tooth fracture varied. Group C had the longest time, followed by group NC; groups ZR, LD, and FC had the shortest time. Similar failure patterns were observed in groups ZR and LD, which were more regular than the pattern in group NC; group FC exhibited the roughest fracture surfaces. Fracture resistance testing combined with a high-speed camera system and fractographic analysis provides a promising modality for studying the biomechanical behaviors of restored teeth. Endocrowns composed of lithium disilicate glass ceramics or resin-based nanoceramics offer alternative restorations for ETMs with extensive coronal loss.

Quantitative Neuroanatomical Measurement on Photogrammetric Model: Validation Study.

To examine and compare the accuracy of measurements obtained from photogrammetric models versus direct measurements taken on dry skulls, with the aim to verify the feasibility of photogrammetry for quantitative analysis in microsurgical neuroanatomy. Two dry human skulls were used. Each was scanned using the dual camera system of a smartphone The selected photos were separately processed using 2 different softwares to create three-dimensional models. Subsequently, 41 anatomical measurements (both linear and curvilinear) based on common anatomical landmarks were taken both directly on dry skulls and on photogrammetric models and compared. Analyzed factors included measurement error, intrarater and interrater reliability, and intermodality agreement. Four photogrammetric models were created. Analysis revealed similar errors when comparing photogrammetric and direct measurements. Measurements from digital models exhibited robust reliability among repeated measures and different observers, supported by very high intraclass correlation coefficient values. Mean measurement difference between Agisoft Metashape software-generated models and direct measurement was 0.01 cm with no systematic bias observed. Conversely, the Metascan app-derived models showed a mean measurement difference of-0.35 cm compared with direct measurement, displaying good agreement for smaller measurements and a systematic proportional bias with increasing measurement size. Two photogrammetric models were validated as quantitative analysis techniques for laboratory neuroanatomical studies, showing acceptable measurement error, high intrarater and interrater reliability, and good to very good agreement compared with direct measurement on dry skulls, replacing expensive and time-consuming methods such as computed tomography scans and neuronavigation systems.

Validation of steps countering recorded with a smartwatch.

Activity trackers are becoming increasingly popular in everyday life, sport and research. Accurate measurement of step counts is essential for understanding an individual's daily activity levels and promoting healthier lifestyles. As the sensor technology has become very sophisticated, such a device can be worn on the body for several hours or days without restricting the subject's daily activities. The aim of this study was to evaluate the accuracy of a commercially available wrist-worn pedometer in accurately recording step counts. A total of twenty-one subjects participated in this comparative study by running at various speeds in a laboratory setting while wearing a Garmin Forerunner 245 Music watch (Garmin Ltd, Schaffhausen, Switzerland) on each wrist. Simultaneously, their steps were recorded by a camera system. The step counts recorded by the pedometer were compared with the counts obtained by the camera system. Descriptive statistical analyses such as Pearson correlation and Bland-Altman plots were used to assess the accuracy and reliability of the pedometer. For most settings the results on the treadmills showed a high correlation between the pedometer and the camera system. On the step machine the results showed a low correlation between the pedometer and the camera system, the pedometer showed a systematic bias by consistently overestimating the number of steps. Bland-Altman plots showed that the results were within the limits of agreement. Although pedometers provide a convenient tool for monitoring step count, this study highlights some limitations and potential inaccuracies of a commercially available pedometer. Our study indicates that the Garmin watch is sufficiently accurate to be used as a measurement tool in a treadmill running environment.

Optimizing 2D-to-3D image conversion for precise flat surface detection using laser triangulation and HSV masking

This study tackles a critical challenge in converting two-dimensional (2D) images into three-dimensional (3D) representations, focusing on the precise detection of flat surfaces. The research utilizes a triangulation method involving laser and camera systems, emphasizing the optimization of laser shooting angles and camera positioning to accurately determine z-coordinates. The methodology employs hue, saturation, and value (HSV) color masking, which has proven superior to traditional red, green, blue (RGB) methods for isolating red line objects. Key findings indicate that the optimal laser angle, β1=70.65°, significantly minimizes root mean square (RMS) error, thereby enhancing the accuracy of 3D imaging. Additionally, the use of three laser lines at different angles enables a more comprehensive detection of z-coordinates by creating multiple reference points across the surface. This arrangement improves the robustness and precision of the 3D reconstruction process, as the intersecting laser lines generate detailed coordinate data that is critical for accurately mapping surface irregularities. These results not only support existing theories in digital feature extraction but also offer a robust framework for practical applications in manufacturing and quality control, particularly in surface defect detection. The study’s innovative approach advances the field of computer vision, providing new insights and methodologies for optimizing image conversion techniques.

Enhancing indoor temperature mapping: High-resolution insights through deep learning and computational fluid dynamics

This paper examines the temperature distribution in a closed, rectangular room equipped with an air conditioning system, employing a computational fluid dynamics model to simulate a virtual thermal camera that captures detailed temperature snapshots. A super-resolution framework enhances the postprocessing of these results. Specifically, convolutional neural networks, trained on simulation data, are used to accurately model temperature fields' high-resolution spatial and temporal evolution. The model demonstrates strong performance by accurately reconstructing temperature profiles from low-resolution inputs obtained from filtering data obtained using high-resolution numerical simulations, with quantitative metrics indicating acceptable accuracy for resolutions reduced by up to 50 times. This effectively aligns with ground truth profiles under various conditions. These results underscore the super-resolution model's potential to transform environmental monitoring in smart buildings and complex structures by generating high-resolution thermal maps from low-resolution cameras or limited sensor input. This approach offers a fast, cost-effective, and reliable method for accurately modeling thermal dynamics within the turbulent flow environments of interior spaces.

Single pixel readout of a MaPMT-based neutron anger camera system with waveform digitizers

In the neutron-sensitive Anger camera, the scintillator is directly coupled to the photodetectors, such as multiplier tubes or silicon-photomultipliers (SiPMs), and the incident position of the neutron is determined by the center of gravity of the photon distribution from the scintillating material. The photodetectors in the neutron Anger camera are typically arranged in arrays with multiple pixel outputs. To reduce the number of readout channels, the front-end electronics often employ analog signal multiplexing methods, such as row-column summation and resistor networks. However, these methods increase channel crosstalk and noise after compression. In this paper, we propose reading out each pixel of the multi-anode photomultiplier tube (MaPMT) individually and digitizing each analog channel via waveform capture. This approach allows direct collection of the signal from each analog channel, and we use the pulse amplitude method to achieve neutron-gamma discrimination for GS20 scintillator. We tested the neutron two-dimensional imaging capability of the anger camera system using a Cf-252 slowed neutron beam in the laboratory. The neutron position resolution was estimated using the knife-edge function method with B4C mask. By combining this with the traditional center-of-gravity method, we achieved a position resolution of 1.2 mm (FWHM).

The Potential of Low-Tech Tools and Artificial Intelligence for Monitoring Blue Carbon in Greenland’s Deep Sea

Arctic environments are changing rapidly. To assess climate change impacts and guide conservation, there is a need to effectively monitor areas of high biodiversity that are difficult to access, such as the deep sea. Greenland (Kalaallit Nunaat), like many remote countries with large deep-sea exclusive economic zones (EEZs), lacks consistent access to the funding and logistics required to maintain advanced and expensive technologies for seafloor exploration. To fill this need, video and camera imaging technologies have been adapted to suit the unique requirements of Arctic environments and the social and economic needs of Greenland. Since 2015, a benthic monitoring program carried out by the Greenland Institute of Natural Resources (GINR) has provided the only large-scale, comprehensive survey in this region, including collection and analysis of photos and GoPro video footage recorded as deep as 1,600 m (Blicher, and Arboe, 2021). In line with the “collect once, use many times” principle, GINR is exploring the versatility of these data, which were originally designated for monitoring and evidence-based management. A potential research avenue for these data is polar blue carbon—the carbon stored and sequestered in ocean habitats—including benthic communities that either live on the seafloor (such as corals and sponges) or are transported there by ocean currents (such as algal detritus). This paper outlines Greenland’s affordable deep-sea technology, based on a towed camera system (Yesson, 2023), and its potential application to rapid, standardized artificial intelligence (AI)-based analysis.

Real-time, accurate, and open source upper-limb musculoskeletal analysis using a single RGBD camera — An exploratory hand-cycling study

Biomechanical biofeedback may enhance rehabilitation and provide clinicians with more objective task evaluation. These feedbacks often rely on expensive motion capture systems (∼$100000), which restricts their widespread use, leading to the development of computer vision-based methods. These methods are subject to large joint angle errors, considering the upper limb, and exclude the scapula and clavicle motion in the analysis. Our open-source approach offers a user-friendly solution for high-fidelity upper-limb kinematics using a single consumer-grade depth-sensing camera (∼$500) and includes semi-automatic skin marker labeling. Real-time biomechanical analysis, ranging from kinematics to muscle force estimation, was conducted on eight participants performing a hand-cycling motion to demonstrate the applicability of our approach on the upper limb. Markers were recorded by the depth-sensing camera and an optoelectronic camera system, considered as a reference. Muscle activity and external load were recorded using eight electromyography sensors and instrumented hand pedals, respectively. Bland-Altman analysis revealed significant agreements in the 3D markers’ positions between the two motion capture methods, with errors averaging 3.3 ± 3.9 mm. The error propagation was low for the biomechanical analysis, with joint angle differences, for example, below 5° when comparing both systems. Biofeedback from the depth-sensing camera was provided at 68 Hz. Our study introduces a novel method for using a depth-sensing camera as a low-cost motion capture solution. Results from healthy participants suggest its potential for accurate kinematic reconstruction and comprehensive upper-limb biomechanical studies. Further investigation is needed to explore its clinical applications in pathological populations.

Development of an energy-efficient cctv camera system for real-time human detection using YOLOv8 model

Human recognition is widely used in variety of fields such as autonomous vehicles, surveillance field, automatons, assisting blind peoples and many more. Many machine learning (ML) and deep learning (DL) algorithms exist for video analysis the main motive of these algorithms is to find human in complicated image. The research presented in this paper focuses on the development of an energy-efficient, smart CCTV camera system for real-time human detection, utilizing the YOLOv8 (You Only Look Once) model. The problem addressed is the need for more advanced, autonomous surveillance systems capable of human detection under various background conditions, overcoming the limitations of traditional CCTV systems, which require constant manual monitoring. The proposed system was trained on the PASCAL VOC 2012 dataset and optimized through hyperparameter tuning, achieving high accuracy and real-time performance. Key results demonstrate that the YOLOv8 model, implemented on the NVIDIA Jetson Nano platform, offers remarkable accuracy, precision, and energy efficiency. It consistently detects human figures in real-time, even in non-ideal conditions like poor lighting or complex backgrounds. This success can be attributed to YOLOv8’s cross-stage partial network (CSPNet) architecture, which enhances its ability to process images quickly and accurately, ensuring it meets the demands of continuous surveillance. The distinguishing features of this system are its energy-efficient design and adaptability to diverse environmental conditions. These characteristics not only solve the challenge of real-time human detection but also make the system a robust and scalable solution for modern security and surveillance applications

Design study of a camera system for Earth observation as a payload on the small satellite ROMEO

AbstractThe importance of a safe food supply has increased due to climate change and its consequences. The number and severity of floods, droughts and plant diseases are rising which causes massive crop failures. Early and precise detection of plant diseases can lower crop failures as it enables early containment. Moreover, it promotes the targeted use of pesticides to protect the biodiversity. Satellite sensors improve the detection of plant diseases by enabling frequent and extensive vegetation observation. Hence, we present the design of a camera system for Earth observation on small satellites with a focus on the detection of plant diseases. The disease detection of sugar beets was chosen as the primary objective due to their importance for the German agriculture. This work is divided into two parts. First, the spectra of sugar beets are analyzed to determine spectral channels for a camera system. Second, a camera system is defined to capture spectral information within the previously defined wavelength ranges. The spectra of healthy and diseased sugar beets were measured in fields for agricultural research in Central Germany using a portable spectroradiometer. The investigated diseases are Cercospora leaf spot and virus yellows, which are among the most important pathogens worldwide. Based on the measured data, the so-called Normalized Difference Sugar Beet Index (NDSBI) is defined, which is a custom index to detect these diseases. We can prove that this index enables more precise detection of these sugar beet diseases than existing indices like the Normalized Difference Vegetation Index (NDVI). The camera system is designed as a payload for the small satellite Research and Observation in Medium Earth Orbit (ROMEO), which is being developed at the University of Stuttgart’s (US) Institute of Space Systems (IRS). The proposed design fulfills the high radiation requirements as well as the system constraints of mass and volume. Three different options for the camera system are developed: two designs for the development at the US, one with lens optics and one with mirror optics, and an adjusted commercial camera system. All defined camera systems permit the measurement of the NDSBI to precisely detect sugar beet diseases.

BGI-YOLO: Background Image-Assisted Object Detection for Stationary Cameras

This paper proposes a method enhancing the accuracy of object detectors by utilizing background images for stationary camera systems. Object detection with stationary cameras is highly valuable across various applications, such as traffic control, crime prevention, and abnormal behavior detection. Deep learning-based object detectors, which are mainly used in such cases, are developed for general purposes and do not take advantage of stationary cameras at all. Previously, cascade-based object detection methods utilizing background have been studied for stationary camera systems. These methods typically consist of two stages: background subtraction followed by object classification. However, their object detection performance is highly dependent on the accuracy of the background subtraction results, and numerous parameters must be adjusted during background subtraction to adapt to varying conditions. This paper proposes an end-to-end object detection method named BGI-YOLO, which uses a background image simply by combining it with an input image before feeding it into the object detection network. In our experiments, the following five methods are compared: three candidate methods of combining input and background images, baseline YOLOv7, and a traditional cascade method. BGI-YOLO, which combines input and background images at image level, showed a detection performance (mAP) improvement compared to baseline YOLOv7, with an increase of 5.6%p on the WITHROBOT S1 dataset and 2.5%p on the LLVIP dataset. In terms of computational cost (GFLOPs), the proposed method showed a slight increase of 0.19% compared to baseline YOLOv7. The experimental results demonstrated that the proposed method is highly effective for improving detection accuracy without increasing computational cost.

Residual Vision Transformer and Adaptive Fusion Autoencoders for Monocular Depth Estimation.

Precision depth estimation plays a key role in many applications, including 3D scene reconstruction, virtual reality, autonomous driving and human-computer interaction. Through recent advancements in deep learning technologies, monocular depth estimation, with its simplicity, has surpassed the traditional stereo camera systems, bringing new possibilities in 3D sensing. In this paper, by using a single camera, we propose an end-to-end supervised monocular depth estimation autoencoder, which contains an encoder with a structure with a mixed convolution neural network and vision transformers and an effective adaptive fusion decoder to obtain high-precision depth maps. In the encoder, we construct a multi-scale feature extractor by mixing residual configurations of vision transformers to enhance both local and global information. In the adaptive fusion decoder, we introduce adaptive fusion modules to effectively merge the features of the encoder and the decoder together. Lastly, the model is trained using a loss function that aligns with human perception to enable it to focus on the depth values of foreground objects. The experimental results demonstrate the effective prediction of the depth map from a single-view color image by the proposed autoencoder, which increases the first accuracy rate about 28% and reduces the root mean square error about 27% compared to an existing method in the NYU dataset.

Augmented reality navigation in orthognathic surgery: Comparative analysis and a paradigm shift.

The emergence of augmented reality (AR) in surgical procedures could significantly enhance accuracy and outcomes, particularly in the complex field of orthognathic surgery. This study compares the effectiveness and accuracy of traditional drilling guides with two AR-based navigation techniques: one utilizing ArUco markers and the other employing small-workspace infrared tracking cameras for a drilling task. Additionally, an alternative AR visualization paradigm for surgical navigation is proposed that eliminates the potential inaccuracies of image detection using headset cameras. Through a series of controlled experiments designed to assess the accuracy of hole placements in surgical scenarios, the performance of each method was evaluated both quantitatively and qualitatively. The findings reveal that the small-workspace infrared tracking camera system is on par with the accuracy of conventional drilling guides, hinting at a promising future where such guides could become obsolete. This technology demonstrates a substantial advantage by circumventing the common issues encountered with traditional tracking systems and surpassing the accuracy of ArUco marker-based navigation. These results underline the potential of this system for enabling more minimally invasive interventions, a crucial step towards enhancing surgical accuracy and, ultimately, patient outcomes. The study resulted in three relevant contributions: first, a new paradigm for AR visualization in the operating room, relying only on exact tracking information to navigate the surgeon is proposed. Second, the comparative analysis marks a critical step forward in the evolution of surgical navigation, paving the way for integrating more sophisticated AR solutions in orthognathic surgery and beyond. Finally, the system with a robotic arm is integrated and the inaccuracies present in a typical human-controlled system areevaluated.