Search

ABSTRACT: Variability in the size of slaughtered chickens remains a longstanding challenge in the standardization of the poultry industry. To address this issue, we present a novel approach that uses volume as a grading metric for chicken carcasses. This innovative method, unexplored in existing studies, employs real-time data capture of moving chicken carcasses on a production line using Kinect v2 depth imaging and 3-D reconstruction technologies. The captured depth images are processed into point clouds followed by 3-D reconstruction. Volume is calculated from the reconstructed models using the surface integration method, and additional 2-D and 3-D features are extracted as input parameters for machine learning models. Multiple regression models were evaluated, with the bagged tree model demonstrating superior performance, achieving an R² value of 0.9988, RMSE of 5.335, and ARE of 2.125%. Furthermore, our method showed remarkable efficiency with an average processing time of less than 1.6 seconds per carcass. These results indicate that our novel approach fills a critical gap in existing automated grading methodologies by offering both accuracy and efficiency. This validates the applicability of depth imaging, 3-D reconstruction, and machine learning for estimating chicken carcass volume with high precision, thereby enabling a more comprehensive, efficient, and reliable chicken carcass grading system.

Considering that structured light depth imagers are difficult to use for precision measurements due to their limited measurement accuracy, we propose an innovative method for correcting structural parameters of structured light depth imagers to reduce the depth measurement error caused by structural parameter errors. For the structured light depth imager, the analytical imaging model is established, and the model of depth error caused by structural parameter errors is established based on the analysis of the depth measurement error analysis. Then, structural parameters are corrected according to the depth measurement error analysis and processing based on experimental depth imaging data of the standard reference plane at the maximum depth. As a result, the corrected analytical imaging model and corrected depth measurement values are obtained. Experimental results have demonstrated the success of this proposed method and its simplicity and convenience. [ABSTRACT FROM AUTHOR]

The accurate depth imaging of piled products provides essential perception for the automated selection of individual objects that require itemized food processing, such as fish, crabs, or fruit. Traditional depth imaging techniques, such as Time-of-Flight and stereoscopy, lack the necessary depth resolution for imaging small items, such as food commodities. Although structured light methods such as laser triangulation have high depth resolution, they depend on conveyor motion for depth scanning. This manuscript introduces an active dual line-laser scanning system for depth imaging static piled items, such as a pile of crabs on a table, eliminating the need for conveyor motion to generate high-resolution 3D images. This advancement benefits robotic perception for loading individual items from a pile for itemized food processing. Leveraging a unique geometrical configuration and laser redundancy, the dual-laser strategy overcomes occlusions while reconstructing a large field of view (FOV) from a long working distance. We achieved a depth reconstruction MSE of 0.3 mm and an STD of 0.5 mm on a symmetrical pyramid stage. The proposed system demonstrates that laser scanners can produce depth maps of complex items, such as piled Chesapeake Blue Crab and White Button mushrooms. This technology enables 3D perception for automated processing lines and offers broad applicability for quality inspection, sorting, and handling of piled products. [ABSTRACT FROM AUTHOR]

Single-photon light detection and ranging (LiDAR) has widespread applications due to the ultra-high sensitivity of the single-photon detector. However, the reconstruction has been plagued by multiple types of noise caused by crosstalk, dark counts, and background light. Artifacts produced by crosstalk are particularly difficult to distinguish directly from the signal for single photon avalanche diode (SPAD) array detector. This work proposes a waveforms-guided depth image estimation method to remove the signal artifacts effectively. Experiments show that a region of interest recognition ratio of 1.02 is achieved at 4.2 km. The study demonstrates that this waveform-guided method has excellent potential for accurate target recognition in a long range.

The time-correlated single-photon-counting (TCSPC) three-dimensional (3D) imaging lidar system has broad application prospects in the field of low-light 3D imaging because of its single-photon detection sensitivity and picoseconds temporal resolution. However, conventional TCSPC systems always limit the echo photon flux to an ultra-low level to obtain high-accuracy depth images, thus needing to spend amounts of acquisition time to accumulate sufficient photon detection events to form a reliable histogram. When the echo photon flux is increased to medium or even high, the data acquisition time can be shortened, but the photon pile-up effect can seriously distort the photon histogram and cause depth errors. To realize high accuracy TCSPC depth imaging with a shorter acquisition time, we propose a high-flux fast photon-counting 3D imaging method based on empirical depth error correction. First, we derive the photon flux estimation formula and calculate the depth error of our photon-counting lidar under different photon fluxes with experimental data. Then, a function correction model between the depth errors and the number of echo photons is established by numerical fitting. Finally, the function correction model is used to correct depth images at high photon flux with different acquisition times. Experimental results show that the empirical error correction method can shorten the image acquisition time by about one order of magnitude while ensuring a moderate accuracy of the depth image. [ABSTRACT FROM AUTHOR]

The CROP95-M18 seismic profile was acquired in 1995 as part of the "CROsta Profonda" project, a regional survey carried out to explore the deeper structures of the Earth's crust in Italy. Although the high energy of the source allowed a high penetration depth, the vintage stacked section showed some areas of low signal-to-noise ratio, due to the acquisition in shallow water environments and related reverberations of surface and guided-waves. To improve the quality of the image, we reprocessed the 2D seismic line with the up-to-date processing techniques by adopting a specific workflow in time domain up to pre-stack time migration. In addition, we created an accurate interval velocity model by combining the coherency inversion technique and reflection tomography. This resulted in depth migration of the uppermost part of the profile (to a depth of 9 km), which was confirmed by the stratigraphy of the Amanda 1bis well. The reprocessing significantly improved the image of the profile and enabled new information on the seismo-stratigraphy to be gleaned from the vintage data. This was particularly important for science and research, as new seismic data with high penetration depths, such as the CROP profiles, are currently difficult to acquire. [ABSTRACT FROM AUTHOR]

Considering that structured light depth imagers are difficult to use for precision measurements due to their limited measurement accuracy, we propose an innovative method for correcting structural parameters of structured light depth imagers to reduce the depth measurement error caused by structural parameter errors. For the structured light depth imager, the analytical imaging model is established, and the model of depth error caused by structural parameter errors is established based on the analysis of the depth measurement error analysis. Then, structural parameters are corrected according to the depth measurement error analysis and processing based on experimental depth imaging data of the standard reference plane at the maximum depth. As a result, the corrected analytical imaging model and corrected depth measurement values are obtained. Experimental results have demonstrated the success of this proposed method and its simplicity and convenience.

The accurate depth imaging of piled products provides essential perception for the automated selection of individual objects that require itemized food processing, such as fish, crabs, or fruit. Traditional depth imaging techniques, such as Time-of-Flight and stereoscopy, lack the necessary depth resolution for imaging small items, such as food commodities. Although structured light methods such as laser triangulation have high depth resolution, they depend on conveyor motion for depth scanning. This manuscript introduces an active dual line-laser scanning system for depth imaging static piled items, such as a pile of crabs on a table, eliminating the need for conveyor motion to generate high-resolution 3D images. This advancement benefits robotic perception for loading individual items from a pile for itemized food processing. Leveraging a unique geometrical configuration and laser redundancy, the dual-laser strategy overcomes occlusions while reconstructing a large field of view (FOV) from a long working distance. We achieved a depth reconstruction MSE of 0.3 mm and an STD of 0.5 mm on a symmetrical pyramid stage. The proposed system demonstrates that laser scanners can produce depth maps of complex items, such as piled Chesapeake Blue Crab and White Button mushrooms. This technology enables 3D perception for automated processing lines and offers broad applicability for quality inspection, sorting, and handling of piled products.

Pig weight measurement is essential for monitoring performance, welfare, and production value. Weight measurement using a scale provides the most accurate results; however, it is time consuming and may increase animal stress. Subjective visual evaluations, even when conducted by an experienced caretaker, lack consistency and accuracy. Optical sensing systems provide alternative methods for estimating pig weight, but studies examining these systems only focus on images taken from stationary cameras. This study fills a gap in existing technology through examining a handheld, portable RGB-D imaging system for estimating pig weight. An Intel RealSense camera collected RGB-D data from finishing pigs at various market weights. 3D point clouds were computed for each pig, and latent features from a 3D generative model were used to predict pig weights using three regression models (SVR, MLP and AdaBoost). These models were compared to two baseline models: median prediction and linear regression using body dimension measurements as predictor variables. Using 10-fold cross validation mean absolute error (MAE) and root-mean-square error (RMSE), all three latent feature models performed better than the median prediction model (MAE = 12.3 kg, RMSE = 16.0 kg) but did not outperform linear regression between weight and girth measurements (MAE = 4.06 kg, RMSE = 4.94 kg). Of the models under consideration, SVR performed best (MAE = 9.25 kg, RMSE = 12.3 kg, mean absolute percentage error = 7.54%) when tested on unseen data. This research is an important step towards developing rapid pig body weight estimation methods from a handheld, portable imaging system by leveraging deep learning feature outputs and depth imaging technology. • A new weight prediction method was developed using mobile, low-cost RGB-D technology. • Method does not restrict the position of the camera relative to pig's body. • Latent features predicted pig weight more accurately than median prediction model. [ABSTRACT FROM AUTHOR]

Seismic imaging is now a well-established method in mineral exploration with many successful case studies. Seismic data are usually imaged in the time domain (post-stack or pre-stack time migration), but recently pre-stack depth imaging has shown clear advantages for irregular/sparse acquisitions and very complex targets. Here, we evaluate the effectiveness of both ray-based and wave-equation-based pre-stack depth imaging methodologies applied to crooked-line 2D seismic reflection profiles. Seismic data were acquired in the Kylylahti mining area in eastern Finland over severely folded, faulted and subvertical Kylylahti structure, and associated mineralization. We performed 3D ray-based imaging, i.e., industry-standard Kirchhoff migration and its improved version (coherency migration, CM), and wave-equation-based migration, i.e., reverse time migration (RTM) using a velocity model built from first-arrival traveltime tomography. Upon comparing the three different migrations against available geological data and models, it appeared that CM provided the least noisy and well-focused image, but failed to image the internal reflectivity of the Kylylahti formation. RTM was the only method that produced geologically plausible reflections inside the Kylylahti formation including a direct image of the previously known shallow massive sulfide mineralization. [ABSTRACT FROM AUTHOR]

The time-correlated single-photon-counting (TCSPC) three-dimensional (3D) imaging lidar system has broad application prospects in the field of low-light 3D imaging because of its single-photon detection sensitivity and picoseconds temporal resolution. However, conventional TCSPC systems always limit the echo photon flux to an ultra-low level to obtain high-accuracy depth images, thus needing to spend amounts of acquisition time to accumulate sufficient photon detection events to form a reliable histogram. When the echo photon flux is increased to medium or even high, the data acquisition time can be shortened, but the photon pile-up effect can seriously distort the photon histogram and cause depth errors. To realize high accuracy TCSPC depth imaging with a shorter acquisition time, we propose a high-flux fast photon-counting 3D imaging method based on empirical depth error correction. First, we derive the photon flux estimation formula and calculate the depth error of our photon-counting lidar under different photon fluxes with experimental data. Then, a function correction model between the depth errors and the number of echo photons is established by numerical fitting. Finally, the function correction model is used to correct depth images at high photon flux with different acquisition times. Experimental results show that the empirical error correction method can shorten the image acquisition time by about one order of magnitude while ensuring a moderate accuracy of the depth image.

The 3D time-of-flight (ToF) cameras have recently received a lot of attention due to their wide range of applications. Despite remarkable advancements in ToF imaging, state-of-the-art ToF cameras are still afflicted by the power hungriness of their illumination sources. To tackle this problem, we exploited existing lighting infrastructure, which ensures the ubiquitous presence of modulated light sources in indoor spaces that serve as opportunity illuminators. We explored the bistatic geometry for passive imaging using the pulse-based ToF approach. Our work is inspired by the recently introduced visible light communication (VLC) or light-fidelity (Li-Fi) infrastructure. VLC allows the infrastructure to provide indoor simultaneous illumination, communication, and sensing (SICS). To this end, we designed a bistatic geometry for the purpose of attaining passive 3D imaging. Such capabilities are achieved by exploiting the pulse shape of the autocorrelation function of real optical signals generated by VLC/Li-Fi modules (e.g., OpenVLC and LiFiMAX). We demonstrated passive imaging by means of matched filtering. In this work, we also studied different sampling strategies in the time-shift domain, including uniform, random, and sparse rulers, which is another step forward toward preserving high depth accuracy with a minimal number of measurements. The proposed methodology achieved successful depth reconstruction with negligible root-mean-square error (RMSE) for the low signal-to-noise ratio (SNR) of the measurements. Parametric models, such as Gaussian and sum of sines, are used to characterize the cross correlation functions and allow for robust parametric depth retrieval from a few measurements. Moreover, we attained a 20-mm worst case error for a target at 25 cm. The experiment proved that the bistatic passive depth reconstruction is feasible. [ABSTRACT FROM AUTHOR]

Background: Half of long-term care (LTC) residents are malnourished, leading to increased hospitalization, mortality, and morbidity, with low quality of life. Current tracking methods are subjective and time-consuming. Objective: This paper presented the automated food imaging and nutrient intake tracking technology designed for LTC. Methods: A needs assessment was conducted with 21 participating staff across 12 LTC and retirement homes. We created 2 simulated LTC intake data sets comprising modified (664/1039, 63.91% plates) and regular (375/1039, 36.09% plates) texture foods. Overhead red-green-blue-depth images of plated foods were acquired, and foods were segmented using a pretrained food segmentation network. We trained a novel convolutional autoencoder food feature extractor network using an augmented UNIMIB2016 food data set. A meal-specific food classifier was appended to the feature extractor and tested on our simulated LTC food intake data sets. Food intake (percentage) was estimated as the differential volume between classified full portion and leftover plates. Results: The needs assessment yielded 13 nutrients of interest, requirement for objectivity and repeatability, and account for real-world environmental constraints. For 12 meal scenarios with up to 15 classes each, the top-1 classification accuracy was 88.9%, with mean intake error of -0.4 (SD 36.7) mL. Nutrient intake estimation by volume was strongly linearly correlated with nutrient estimates from mass (r2=0.92-0.99), with good agreement between methods (s=-2.7 to -0.01; 0 within each of the limits of agreement). Conclusions: The automated food imaging and nutrient intake tracking approach is a deep learning-powered computational nutrient sensing system that appears to be feasible (validated accuracy against gold-standard weighed food method, positive end user engagement) and may provide a novel means for more accurate and objective tracking of LTC residents' food intake to support and prevent malnutrition tracking strategies. [ABSTRACT FROM AUTHOR]

We present TIMo (Time-of-flight Indoor Monitoring), a dataset for video-based monitoring of indoor spaces captured using a time-of-flight (ToF) camera. The resulting depth videos feature people performing a set of different predefined actions, for which we provide detailed annotations. Person detection for people counting and anomaly detection are the two targeted applications. Most existing surveillance video datasets provide either grayscale or RGB videos. Depth information, on the other hand, is still a rarity in this class of datasets in spite of being popular and much more common in other research fields within computer vision. Our dataset addresses this gap in the landscape of surveillance video datasets. The recordings took place at two different locations with the ToF camera set up either in a top-down or a tilted perspective on the scene. Moreover, we provide experimental evaluation results from baseline algorithms. [ABSTRACT FROM AUTHOR]

• Depth imaging through high levels of atmospheric obscurants with limited statistical frames. • A modified Gamma distribution model and multi-scale superpixels are used to solve signal photon-starved but noisy data. • Efficient noise removal and improvement of target integrity are achieved by using the spatial similarity features and the edge information. • Significantly improve target integrity of the reconstructed depth image under such extreme conditions. • Reduce the statistical frame requirement from several thousand to several hundred. The limited statistical frame data and substantial backscattering interference from atmospheric obscurants result in a photon-starved regime, which seriously limits the depth imaging capability of array Gm-APD LiDAR in strong scattering environments. Here, we propose a depth image estimation algorithm through atmospheric obscurants that can significantly improve target integrity when signal photons are scarce. At the signal level, based on the established array Gm-APD LiDAR smoke echo model, this algorithm enhances the number of signal photons by constructing multi-scale superpixels. At the image level, efficient noise removal and improvement of target integrity are achieved by using the spatial similarity features and the edge information of reconstructed images at different scales. It has been successfully demonstrated in different attenuation lengths and atmospheric obscurants. Especially when the visibility is 1.7 km, we acquire depth images through dense fog equivalent to 1.5 attenuation lengths at distances of 1.4 km by using only 800 statistical frames data. This study has great potential for rapid depth imaging of dynamic targets under extreme weather conditions. [ABSTRACT FROM AUTHOR]

A laser scanning confocal microscope (LSCM) is an effective scientific instrument for studying sub-micron structures, and it has been widely used in the field of biological detection. However, the illumination depth of LSCMs is limited due to the optical aberrations introduced by living biological tissue, which acts as an optical medium with a non-uniform refractive index, resulting in a significant dispersion of the focus of LSCM illumination light and, hence, a loss in the resolution of the image. In this study, to minimize the effect of optical aberrations, an image-based adaptive optics technology using an optimized stochastic parallel gradient descent (SPGD) algorithm with an adaptive coefficient is applied to the optical path of an LSCM system. The effectiveness of the proposed aberration correction approach is experimentally evaluated in the LSCM system. The results illustrate that the proposed adaptive optics system with an adaptive coefficient SPGD algorithm can effectively reduce the interference caused by aberrations during depth imaging. [ABSTRACT FROM AUTHOR]

Depth imaging is very important for many emerging technologies, such as artificial intelligence, driverless vehicles and facial recognition. However, all these applications demand compact and low-power systems that are beyond the capabilities of most state-of-art depth cameras. Recently, metasurface-based depth imaging that exploits point spread function (PSF) engineering has been demonstrated to be miniaturized and single shot without requiring active illumination or multiple viewpoint exposures. A pair of spatially adjacent metalenses with an extended depth-of-field (EDOF) PSF and a depth-sensitive double-helix PSF (DH-PSF) were used, using the former metalens to reconstruct clear images of each depth and the latter to accurately estimate depth. However, due to these two metalenses being non-coaxial, parallax in capturing scenes is inevitable, which would limit the depth precision and field of view. In this work, a bifunctional reconfigurable metalens for 3D depth imaging was proposed by dynamically switching between EDOF-PSF and DH-PSF. Specifically, a polarization-independent metalens working at 1550 nm with a compact 1 mm2 aperture was realized, which can generate a focused accelerating beam and a focused rotating beam at the phase transition of crystalline and amorphous Ge2Sb2Te5 (GST), respectively. Combined with the deconvolution algorithm, we demonstrated the good capabilities of scene reconstruction and depth imaging using a theoretical simulation and achieved a depth measurement error of only 3.42%. [ABSTRACT FROM AUTHOR]

The exploration of biogenic gas in the Adriatic Sea is highly developed and the investigation of the residual potential and located complex areas, requires an improvement in the quality of available data. This paper illustrates a 2-year seismic reprocessing program intended to assess the residual gas potential in this still productive hydrocarbon area by leveraging on integrated geological and geophysical workflows, modern processing techniques and High Parallel Computing capabilities. Legacy or vintage data sets can usually be re-examined using new processing technologies and the re-processing of available data is the only solution to re-evaluate the residual potential of developed-explored areas. [ABSTRACT FROM AUTHOR]

Confocal microscopes can provide depth-based imaging which can be useful in detecting skin cancers in early stages. However, confocal microscopes are very costly and are not easily available. To solve this problem, a study on the optical and electrical components of discarded DVD players have been done to prove their usability in a miniature, and cost-effective scanning confocal microscope. The Optical pickup of DVD players is studied to be used as the light source and detector for the microscope. A single mode fiber is used to carry the laser light from the source to the sample and back. The single mode fiber also works as a pinhole, satisfying the confocal condition of the microscope. Electrical components of DVD players have been used for developing a scanner stage. Several experiments have been performed which prove the workability of the setup. [ABSTRACT FROM AUTHOR]

To achieve a simple and less invasive registration procedure in computer-assisted orthopaedic surgery, we propose an automatic, markerless registration and tracking method based on depth imaging and deep learning. A depth camera is used to continuously capture RGB and depth images of the exposed bone during surgery, and deep neural networks are trained to first localise the surgical target using the RGB image, then segment the target area of the corresponding depth image, from which the surface geometry of the target bone can be extracted. The extracted surface is then compared to a pre-operative model of the same bone for registration. This process can be performed dynamically during the procedure at a rate of 5-6 Hz, without any need for surgeon intervention or invasive optical markers. Ex vivo registration experiments were performed on a cadaveric knee, and accuracy measurements against an optically tracked ground truth resulted in a mean translational error of 2.74 mm and a mean rotational error of 6.66°. Our results are the first to describe a promising new way to achieve automatic markerless registration and tracking in computer-assisted orthopaedic surgery, demonstrating that truly seamless registration and tracking of the limb is within reach. Our method reduces invasiveness by removing the need for percutaneous markers. The surgeon is also exempted from inserting markers and collecting registration points manually, which contributes to a more efficient surgical workflow and shorter procedure time in the operating room.

Seismic imaging is now a well-established method in mineral exploration with many successful case studies. Seismic data are usually imaged in the time domain (post-stack or pre-stack time migration), but recently pre-stack depth imaging has shown clear advantages for irregular/sparse acquisitions and very complex targets. Here, we evaluate the effectiveness of both ray-based and wave-equation-based pre-stack depth imaging methodologies applied to crooked-line 2D seismic reflection profiles. Seismic data were acquired in the Kylylahti mining area in eastern Finland over severely folded, faulted and subvertical Kylylahti structure, and associated mineralization. We performed 3D ray-based imaging, i.e., industry-standard Kirchhoff migration and its improved version (coherency migration, CM), and wave-equation-based migration, i.e., reverse time migration (RTM) using a velocity model built from first-arrival traveltime tomography. Upon comparing the three different migrations against available geological data and models, it appeared that CM provided the least noisy and well-focused image, but failed to image the internal reflectivity of the Kylylahti formation. RTM was the only method that produced geologically plausible reflections inside the Kylylahti formation including a direct image of the previously known shallow massive sulfide mineralization.

The methodology presented in this paper covers the topic of automatic detection of humans based on two types of images that do not rely on the visible light spectrum, namely on thermal and depth images. Various scenarios are considered with the use of deep neural networks being extensions of Faster R-CNN models. Apart from detecting people, independently, with the use of depth and thermal images, we proposed two data fusion methods. The first approach is the early fusion method with a 2-channel compound input. As it turned out, its performance surpassed that of all other methods tested. However, this approach requires that the model be trained on a dataset containing both types of spatially and temporally synchronized imaging sources. If such a training environment cannot be setup or if the specified dataset is not sufficiently large, we recommend the late fusion scenario, i.e. the other approach explored in this paper. Late fusion models can be trained with single-source data. We introduce the dual-NMS method for fusing the depth and thermal imaging approaches, as its results are better than those achieved by the common NMS

As depth cameras have gotten smaller, more affordable, and more precise, they have also emerged as a promising sensor in ubiquitous systems, particularly for detecting objects, scenes, and persons. This article sets out to systematically evaluate how suitable depth data can be for picking up users’ respiration, from small distance changes across the torso over time. We contribute a large public dataset of depth data over time from 19 persons taken in a large variety of circumstances. On this data, we evaluate and compare different state-of-the-art methods and show that their individual performance significantly depends on a range of conditions and parameters. We investigate the influence of the observed torso region (e.g., the chest), the user posture and activity, the distance to the depth camera, the respiratory rate, the gender, and user specific peculiarities. Best results hereby are obtained from the chest whereas the abdomen is least suited for detecting the user’s breathing. In terms of accuracy and signal quality, the largest differences are observed on different user postures and activities. All methods can maintain a mean accuracy of above 92% when users are sitting, but half of the observed methods only achieve a mean accuracy of 51% while standing. When users are standing and additionally move their arms in front of their upper body, mean accuracy values between the worst and best performing methods range from 21 to 87%. Increasing the distance to the depth camera furthermore results in lower signal quality and decreased accuracy on all methods. Optimal results can be obtained at distances of 1–2 m. Different users have been found to deliver varying qualities of breathing signals. Causes range from clothing, over long hair, to movement. Other parameters have shown to play a minor role in the detection of users’ breathing.

Variability in the size of slaughtered chickens remains a longstanding challenge in the standardization of the poultry industry. To address this issue, we present a novel approach that uses volume as a grading metric for chicken carcasses. This innovative method, unexplored in existing studies, employs real-time data capture of moving chicken carcasses on a production line using Kinect v2 depth imaging and 3-D reconstruction technologies. The captured depth images are processed into point clouds followed by 3-D reconstruction. Volume is calculated from the reconstructed models using the surface integration method, and additional 2-D and 3-D features are extracted as input parameters for machine learning models. Multiple regression models were evaluated, with the bagged tree model demonstrating superior performance, achieving an R² value of 0.9988, RMSE of 5.335, and ARE of 2.125%. Furthermore, our method showed remarkable efficiency with an average processing time of less than 1.6 seconds per carcass. These results indicate that our novel approach fills a critical gap in existing automated grading methodologies by offering both accuracy and efficiency. This validates the applicability of depth imaging, 3-D reconstruction, and machine learning for estimating chicken carcass volume with high precision, thereby enabling a more comprehensive, efficient, and reliable chicken carcass grading system., (Copyright © 2024 The Authors. Published by Elsevier Inc. All rights reserved.)

The methodology presented in this paper covers the topic of automatic detection of humans based on two types of images that do not rely on the visible light spectrum, namely on thermal and depth images. Various scenarios are considered with the use of deep neural networks being extensions of Faster R-CNN models. Apart from detecting people, independently, with the use of depth and thermal images, we proposed two data fusion methods. The first approach is the early fusion method with a 2-channel compound input. As it turned out, its performance surpassed that of all other methods tested. However, this approach requires that the model be trained on a dataset containing both types of spatially and temporally synchronized imaging sources. If such a training environment cannot be setup or if the specified dataset is not sufficiently large, we recommend the late fusion scenario, i.e. the other approach explored in this paper. Late fusion models can be trained with single-source data. We introduce the dual-NMS method for fusing the depth and thermal imaging approaches, as its results are better than those achieved by the common NMS. [ABSTRACT FROM AUTHOR]

In order to obtain the respiratory condition unobtrusively and comfortably, a non-contact method based on the commercial depth camera Realsense SR300 was proposed to extract respiratory information from depth data. In this paper, a respiratory region detecting algorithm which is mainly based on the morphological method was proposed to obtain the region of interest (ROI) with the depth images. The proposed algorithm contains four steps: body edge extraction, noise reduction, “image skeleton” extraction, and respiratory region estimation. As a result, the respiratory waveform can be derived from the depth data in the ROI. For validation, experiments were carried out to verify the feasibility of obtaining the respiratory information with this approach. In consideration of different application scenarios, 20 kinds of conditions were designed and applied for the experiments. The respiratory rate extracted from the depth waveform can be calculated, and the accuracy achieved was 95.20% for all data while utilizing polysomnography thorax effort signal as gold standard. Through the Bland–Altman analysis, it represented that the proposed system had a good agreement ( $r^{2} = 0.88$ ) with the gold standard. In addition, the performances of the system in the 20 different conditions were analyzed by statistics, and the results showed that the system has good adaptability and robustness for different conditions. In conclusion, the proposed algorithm can fit different scenarios, and this paper provides a novel option for extracting the physiological information with depth data.

Accurately measuring body weight (BW) is of great importance for cow-calf producers in the United States, which can lead to better marketing and nutrient management decisions. Traditionally, BW has been recorded using a walk-over scale or approximated by visual observation. Recently, technology has come out that allows BW prediction using three-dimensional (3D) images. The objective of this study was to determine the efficacy of 3D imaging technology as a method to predict BW of beef heifers. Dorsal 3D images were collected on 69 Red Angus/Simmental yearling heifers. Heifers were approximately 12 mo of age that ranged from 282.1 to 440.0 kg of BW. For image collection, a depth camera (Kinect Azure, Microsoft) was placed above the animals, approximately 3 to 4 meters above the floor surface within the alley of a working facility. The image collection software was written in Python (version 3.8) programming language. For a high-quality image to be recorded, heifers needed to be handled in small groups and kept separate. Heifers also needed to have all four feet on the ground and no interference from other animals in the image. Corresponding scale-measured BW were also recorded for each animal concurrent with depth images collection. Heifers were limit fed and weighed in the morning before feeding for both scale measured BW and image collection. The depth images were analyzed using a customized program written in MATLAB (R2022a). Heifer dorsal projected volume was estimated by adding height pixel values that formed the heifer's dorsal area, excluding the head region. Body weights were regressed from projected heifer body volumes extracted from the depth images, and then metabolic BW (MBW) was calculated using the 0.75 power adjustment of BW. Performance of the prediction model was analyzed using the coefficient of determination (R2) obtained from the regression of projected volume and actual scale BW and the standard error of the mean (SEM) of the predicted BW. Data were analyzed using the PROC REG and PROC CORR procedures in SAS (v9.4). Prediction of BW using image-extracted dorsal projected volume produced an R 2 = 0.89 and SEM = 3.3 kg. When analyzing the correlation of volume to BW, the Pearson correlation coefficient of r = 0.94 (P < 0.0001) was obtained. Regression of projected volume to MBW produced an R 2 = 0.89 and SEM = 0.57 kg. Results show that this model demonstrates the ability to accurately predict BW using depth images projected dorsal volumes on yearling heifers. These results illustrate that 3D images can be used to accurately measure BW of growing heifers that may enhance the ability of cow-calf producers to pro-actively manage grazing livestock. [ABSTRACT FROM AUTHOR]

A laser scanning confocal microscope (LSCM) is an effective scientific instrument for studying sub-micron structures, and it has been widely used in the field of biological detection. However, the illumination depth of LSCMs is limited due to the optical aberrations introduced by living biological tissue, which acts as an optical medium with a non-uniform refractive index, resulting in a significant dispersion of the focus of LSCM illumination light and, hence, a loss in the resolution of the image. In this study, to minimize the effect of optical aberrations, an image-based adaptive optics technology using an optimized stochastic parallel gradient descent (SPGD) algorithm with an adaptive coefficient is applied to the optical path of an LSCM system. The effectiveness of the proposed aberration correction approach is experimentally evaluated in the LSCM system. The results illustrate that the proposed adaptive optics system with an adaptive coefficient SPGD algorithm can effectively reduce the interference caused by aberrations during depth imaging.

We present TIMo (Time-of-flight Indoor Monitoring), a dataset for video-based monitoring of indoor spaces captured using a time-of-flight (ToF) camera. The resulting depth videos feature people performing a set of different predefined actions, for which we provide detailed annotations. Person detection for people counting and anomaly detection are the two targeted applications. Most existing surveillance video datasets provide either grayscale or RGB videos. Depth information, on the other hand, is still a rarity in this class of datasets in spite of being popular and much more common in other research fields within computer vision. Our dataset addresses this gap in the landscape of surveillance video datasets. The recordings took place at two different locations with the ToF camera set up either in a top-down or a tilted perspective on the scene. Moreover, we provide experimental evaluation results from baseline algorithms.

Depth imaging is very important for many emerging technologies, such as artificial intelligence, driverless vehicles and facial recognition. However, all these applications demand compact and low-power systems that are beyond the capabilities of most state-of-art depth cameras. Recently, metasurface-based depth imaging that exploits point spread function (PSF) engineering has been demonstrated to be miniaturized and single shot without requiring active illumination or multiple viewpoint exposures. A pair of spatially adjacent metalenses with an extended depth-of-field (EDOF) PSF and a depth-sensitive double-helix PSF (DH-PSF) were used, using the former metalens to reconstruct clear images of each depth and the latter to accurately estimate depth. However, due to these two metalenses being non-coaxial, parallax in capturing scenes is inevitable, which would limit the depth precision and field of view. In this work, a bifunctional reconfigurable metalens for 3D depth imaging was proposed by dynamically switching between EDOF-PSF and DH-PSF. Specifically, a polarization-independent metalens working at 1550 nm with a compact 1 mm2 aperture was realized, which can generate a focused accelerating beam and a focused rotating beam at the phase transition of crystalline and amorphous Ge2Sb2Te5 (GST), respectively. Combined with the deconvolution algorithm, we demonstrated the good capabilities of scene reconstruction and depth imaging using a theoretical simulation and achieved a depth measurement error of only 3.42%.

A case history is presented describing the revision of the 3D seismic volume for an onshore hydrocarbon field in Italy. One of the key elements for the success of the project was the contribution of the geological information to seismic data processing and imaging, which was deemed since the beginning as an indispensable guide to drive and constraint the seismic reprocessing. In particular, two critical steps in which the integration between geology and geophysics played a relevant role are discussed here: the estimation of static corrections for data preconditioning before imaging, and the velocity model building for depth imaging. The paper includes also a proposal for a stratigraphic revision of the allochthonous sequence of the Val d’Agri area. The outcomes of the project comprised the release of a reprocessed seismic volume in time domain, the creation of the first Pre-Stack Depth Migrated volume of the field, together with the revision of some stratigraphic boundaries, and created the conditions for a revision of the geological model of the field. The project lasted approximately two years and was executed by a multidisciplinary group comprising geophysicists, geologists, sedimentologists, and structuralists. The collaboration with Shell, joint venture partner, was also an important factor for the success of the project. [ABSTRACT FROM AUTHOR]

Rapid advances in medical imaging technology, particularly the development of optical systems with non-linear imaging modalities, are boosting deep tissue imaging. The development of reliable standards and phantoms is critical for validation and optimization of these cutting-edge imaging techniques. We aim to design and fabricate flexible, multi-layered hydrogel-based optical standards and evaluate advanced optical imaging techniques at depth. Standards were made using a robust double-network hydrogel matrix consisting of agarose and polyacrylamide. The materials generated ranged from single layers to more complex constructs consisting of up to seven layers, with modality-specific markers embedded between the layers. These standards proved useful in the determination of the axial scaling factor for light microscopy and allowed for depth evaluation for different imaging modalities (conventional one-photon excitation fluorescence imaging, two-photon excitation fluorescence imaging, second harmonic generation imaging, and coherent anti-Stokes Raman scattering) achieving actual depths of 1550, 1550, 1240, and 1240 μm , respectively. Once fabricated, the phantoms were found to be stable for many months. The ability to image at depth, the phantom's robustness and flexible layered structure, and the ready incorporation of "optical markers" make these ideal depth standards for the validation of a variety of imaging modalities. [ABSTRACT FROM AUTHOR]

This paper proposes a novel method to filter out the false alarm of LiDAR system by using the temporal correlation of target reflected photons. Because of the inevitable noise, which is due to background light and dark counts of the detector, the depth imaging of LiDAR system exists a large estimation error. Our method combines the Poisson statistical model with the different distribution feature of signal and noise in the time axis. Due to selecting a proper threshold, our method can effectively filter out the false alarm of system and use the ToFs of detected signal photons to rebuild the depth image of the scene. The experimental results reveal that by our method it can fast distinguish the distance between two close objects, which is confused due to the high background noise, and acquire the accurate depth image of the scene. Our method need not increase the complexity of the system and is useful in power-limited depth imaging.

Depth imaging has, through recent technological advances, become ubiquitous as products become smaller, more affordable, and more precise. Depth cameras have also emerged as a promising modality for activity recognition as they allow detection of users’ body joints and postures. Increased resolutions have now enabled a novel use of depth cameras that facilitate more fine-grained activity descriptors: The remote detection of a person’s breathing by picking up the small distance changes from the user’s chest over time. We propose in this work a novel method to model chest elevation to robustly monitor a user’s respiration, whenever users are sitting or standing, and facing the camera. The method is robust to users occasionally blocking their torso region and is able to provide meaningful breathing features to allow classification in activity recognition tasks. We illustrate that with this method, with specific activities such as paced-breathing meditating, performing breathing exercises, or post-exercise recovery, our model delivers a breathing accuracy that matches that of a commercial respiration chest monitor belt. Results show that the breathing rate can be detected with our method at an accuracy of 92 to 97% from a distance of two metres, outperforming state-of-the-art depth imagining methods especially for non-sedentary persons, and allowing separation of activities in respiration-derived features space.

A CMOS time-of-flight (TOF) image sensor with in-pixel background light (BGL) cancellation and demodulation lock-in pixel is developed for both indoor and outdoor depth imaging applications. The proposed polarity switching integration technique using p+/n-well diode realizes the in-pixel BGL cancellation up to 180 klx. In addition, the developed phase-shift readout (PSR) effectively reduces the inherent correlated noise and column fixed-pattern noise (FPN) as well with sensitivity improvement. A prototyped pulse-modulation-based TOF sensing chip with a $64 \times 64$ pixel array, $20\mu \text {m} \times 20\mu \text{m}$ pixel pitch, and 33% fill factor has been fabricated in TSMC standard 0.18- $\mu \text{m}$ CMOS process and verified. The achieved depth sensing capability ranges from 0.75 to 7.5 m with a linearity error below 1.1%, and the measured relative precision is 4.2% at a 7.5-m target distance. [ABSTRACT FROM AUTHOR]

Highlights • Propose a novel Region Ensemble Network for end-to-end hand pose estimation. • Present several good practices of ConvNets for hand pose estimation. • The Proposed method outperforms state-of-the-arts on hand and human pose datasets. 3D hand pose estimation is an important and challenging problem for human-computer interaction. Recently convolutional networks (ConvNet) with sophisticated design have been employed to address it, but the improvement is not so significant. To exploit good practice and promote the performance for hand pose estimation, we propose a Region Ensemble Network (REN) for directly 3D coordinate regression. It first partitions the last convolutional outputs of ConvNet into several grid regions. Results from separate fully-connected (FC) regressors on each regions are integrated by another FC layer to perform estimation. By exploitation of several training strategies including data augmentation and smooth L 1 loss, REN significantly improves the performance of ConvNet for hand pose estimation. Experiments demonstrate that our approach achieves strong performance on par or better than state-of-the-art algorithms on three public hand pose datasets. We also experiment our methods on fingertip detection and human pose datasets and obtain state-of-the-art accuracy. [ABSTRACT FROM AUTHOR]

Robotic ultrasound systems have turned into clinical use over the past few decades, increasing precision and quality of medical operations. In this paper, we propose a fully automatic scanning system for three-dimensional (3-D) ultrasound imaging. A depth camera was first used to obtain the depth data and color data of the tissue surface. Based on the depth image, the 3-D contour of the tissue was rendered and the scan path of ultrasound probe was automatically planned. Following the scan path, a 3-D translating device drove the probe to move on the tissue surface. Simultaneously, the B-scans and their positional information were recorded for subsequent volume reconstruction. In order to stop the scanning process when the pressure on the skin exceeded a preset threshold, two force sensors were attached to the front side of the probe for force measurement. In vitro and in vivo experiments were conducted for assessing the performance of the proposed system. Quantitative results show that the error of volume measurement was less than 1%, indicating that the system is capable of automatic ultrasound scanning and 3-D imaging. It is expected that the proposed system can be well used in clinical practices. [ABSTRACT FROM PUBLISHER]

• A new operation mode of time-to-digital converters to separate the detection of the minimum distance and the depth of field in time-of-flight measurements. • The mode achieves a long detection range with high resolution in a small pixel area. • The mode reduces the power consumption of the readout integrated circuit. A novel time-to-digital converter (TDC) operation mode, called double-frame-multiplexing (DFM), is proposed for time-of-flight (TOF) measurements. It is based on the separate detection of the minimum distance and depth of the measured targets, which reduces the power consumption of in-pixel TDCs and expands the maximum detection distance under high resolution. A 64 × 64 direct TOF image sensor chip is fabricated in a 0.18 μm CMOS process. Driven by a 125 MHz multi-phase clock and a 50 MHz data readout clock, the chip achieves a maximum frame rate of 20 kframe/s, a maximum detection distance of 750 m (5 μs), and a maximum depth of field of 3 m (20 ns) with a temporal resolution of 1 ns. The DFM mode reduces the average power consumption to 60 mW without consideration of SPADs. The worst-case measured differential nonlinearity is −0.61/+0.69 least significant bit (LSB) and the maximum integral nonlinearity is −1.06/+0.66 LSB. [ABSTRACT FROM AUTHOR]