Three-dimensional Reconstruction Method Research Articles

Interactive 3D reconstruction method of fuzzy static images in social media

Abstract Because the traditional social media fuzzy static image interactive three-dimensional (3D) reconstruction method has the problem of poor reconstruction completeness and long reconstruction time, the social media fuzzy static image interactive 3D reconstruction method is proposed. For preprocessing the fuzzy static image of social media, the Harris corner detection method is used to extract the feature points of the preprocessed fuzzy static image of social media. According to the extraction results, the parameter estimation algorithm of contrast divergence is used to learn the restricted Boltzmann machine (RBM) network model, and the RBM network model is divided into input, output, and hidden layers. By combining the RBM-based joint dictionary learning method and a sparse representation model, an interactive 3D reconstruction of fuzzy static images in social media is achieved. Experimental results based on the CAD software show that the proposed method has a reconstruction completeness of above 95% and the reconstruction time is less than 15 s, improving the completeness and efficiency of the reconstruction, effectively reconstructing the fuzzy static images in social media, and increasing the sense of reality of social media images.

ASPPMVSNet: A high‐receptive‐field multiview stereo network for dense three‐dimensional reconstruction

The learning-based multiview stereo (MVS) methods for three-dimensional (3D) reconstruction generally use 3D volumes for depth inference. The quality of the reconstructed depth maps and the corresponding point clouds is directly influenced by the spatial resolution of the 3D volume. Consequently, these methods produce point clouds with sparse local regions because of the lack of the memory required to encode a high volume of information. Here, we apply the atrous spatial pyramid pooling (ASPP) module in MVS methods to obtain dense feature maps with multiscale, long-range, contextual information using high receptive fields. For a given 3D volume with the same spatial resolution as that in the MVS methods, the dense feature maps from the ASPP module encoded with superior information can produce dense point clouds without a high memory footprint. Furthermore, we propose a 3D loss for training the MVS networks, which improves the predicted depth values by 24.44%. The ASPP module provides state-of-the-art qualitative results by constructing relatively dense point clouds, which improves the DTU MVS dataset benchmarks by 2.25% compared with those achieved in the previous MVS methods.

Three-Dimensional Reconstruction Method for Bionic Compound-Eye System Based on MVSNet Network

In practical scenarios, when shooting conditions are limited, high efficiency of image shooting and success rate of 3D reconstruction are required. To achieve the application of bionic compound eyes in small portable devices for 3D reconstruction, auto-navigation, and obstacle avoidance, a deep learning method of 3D reconstruction using a bionic compound-eye system with partial-overlap fields was studied. We used the system to capture images of the target scene, then restored the camera parameter matrix by solving the PnP problem. Considering the unique characteristics of the system, we designed a neural network based on the MVSNet network structure, named CES-MVSNet. We fed the captured image and camera parameters to the trained deep neural network, which can generate 3D reconstruction results with good integrity and precision. We used the traditional multi-view geometric method and neural networks for 3D reconstruction, and the difference between the effects of the two methods was analyzed. The efficiency and reliability of using the bionic compound-eye system for 3D reconstruction are proved.

Design of a multi-sensor information acquisition system for mannequin reconstruction and human body size measurement under clothes

Although the measurement of human sizes and the reconstruction of mannequins have been extensively studied, there has been little research on systems applicable for measurement and reconstruction under clothes. This work presents a method for three-dimensional (3D) mannequin reconstruction and human body size measurement based on information fusion of multiple sensors. For this purpose, a multi-sensor information acquisition system (MIAS) is developed allowing one to acquire 11 key size data of the human body under clothes. The MIAS is based on the proposed parametric mannequin method and the Laplace mesh deformation technique, thus allowing one to reconstruct the mannequin conforming to the MIAS measurements. All mannequins reconstructed by this method have the same topology, which will greatly facilitate subsequent processing, such as garment design, customization and virtual fitting. The results of the reconstruction experiments show that our system can reconstruct a mannequin that matches the size of the feature parts very well. In addition, performed comparison experiments with manual and laser measurements shows that the measurement results of our system are close to those of manual and laser measurements. This provides a viable method for measuring key body sizes under clothes and reconstructing the corresponding 3D mannequin.

Research on Reconstruction of Basketball Training Action Trajectory Based on Improved K-Means Clustering Algorithm

The automatic capture and analysis of basketball game movements can guide basketball training and provide an effective method for improving the efficiency of basketball training. This paper introduces the research status of clustering methods in the field of trajectory data mining and reconstruction in detail. By analyzing the trajectory data under the constraints of the road network, the spatiotemporal characteristics of the existing trajectory clustering methods, and the deficiencies of the existing trajectory clustering methods, a new trajectory clustering method based on trajectory segmentation and spatiotemporal similarity measurement is implemented. A motion capture and reconstruction method for basketball training based on visual image K-means clustering algorithm is proposed. Multiresolution frame scanning technology is used to collect machine images of basketball training movements, and edge contour processing is performed on the collected high-resolution basketball training movement images. Feature detection uses the three-dimensional model reconstruction method to segment the basketball training action area and combines the irregular triangle network model to realize the machine vision block template matching processing of basketball training actions and capture the basketball training action in the Gaussian fuzzy affine space. In time and feature extraction, wavelet lifting technology is used to identify the ambiguity of basketball training movements, image enhancement technology is used to improve the resolution and adaptability of basketball training movement capture, and machine vision image processing methods are used to achieve basketball training movement capture optimization. The simulation results show that the method has better adaptability and higher feature recognition ability for basketball training motion capture and improves the feature extraction and adaptive capture reconstruction ability of basketball training motion.

The correlation algorithm based on jittering structured light with multiple reflection imaging

As a three-dimensional reconstruction method, the structured light formed by multiple reflection of neutral beam splitter mirror can effectively solve the problem caused by high reflectivity. In order to obtain more stripes, the jittering mechanism is used in the system. In addition, the linear algorithm can not make full use of the information between different structured light, which will produce a large error and inaccurate result. To solve this problem, we propose a correlation algorithm. In this algorithm, we use two structured light as the independent variables to obtain spatial information, and a related items is calculated to take full advantage of the information between different structured light. Experimental results show that the overall measurement error is reduced by 20-30% and the repetition accuracy of the system reaches 0.02 mm.

Three-dimensional measurement and analysis of morphological parameters of the uterus in infertile women.

To determine differences in endometrial cavity anteroposterior diameter, thickness, volume, and diameter lines of uterine body and thickness, and volume of upper, middle, and lower regions of the endometrium in infertile women using a new method for three-dimensional (3D) reconstruction based on two-dimensional (2D) ultrasound images. This retrospective cross-sectional study included a total of 81 infertile women, who underwent 2D ultrasound standard examination. We created 3D models of the uterine body, endometrial cavity, and endometrium based on 2D ultrasound images. The parameters that were measured and analyzed in a 3D plane included volume and diameter lines of endometrial cavity, surface area, thickness, volume, and diameter lines of uterine body, and surface area, thickness, and volume of upper, middle, and lower region of the endometrium. These parameters were used for comparisons between normal and arcuate uterus, between non-pregnant and pregnant infertile women, and between nulliparous and multiparous infertile women. The differences between the different regions of the endometrium and the correlations between age and the parameters were also determined in this study. Endometrial cavity length, and middle and lower regions of the endometrial volume in the normal uterus were 39.63±7.61 mm, 1,307.92±1,034.40 mm3, and 653.98±460.41 mm3, respectively. For arcuate uterus, these parameters were 32.96±4.69 mm, 539.89±298.94 mm3, and 347.90±129.61 mm3, respectively. The parameters were significantly higher in normal uterus compared with arcuate uterus (P=0.000, 0.001, and 0.006, respectively). Upper, middle, and lower regions of endometrial thickness in normal uterus were 7.79±3.26, 8.18±3.33, and 6.41±2.60 mm, respectively. Both upper and middle regions of endometrial thickness were significantly greater than the lower regions of endometrial thickness with P=0.009 and P=0.001, respectively. Correlation analysis revealed that age positively correlated with volume of upper endometrial regions (r=0.274, P=0.028). This study provides references for the volume and thickness of the endometrium in the different anatomical regions of normal and arcuate uterus. Age mainly affects the upper region of the endometrium. The 3D measurement provides a precise way to quantify the morphological parameters of gynecological diseases.

Three-Dimensional Geometry Reconstruction Method for Slowly Rotating Space Targets Utilizing ISAR Image Sequence

For a slowly rotating space target (SRST) with a fixed axis, traditional 3D geometry reconstruction methods become invalid as the projection vectors cannot be formed without accurate target rotational parameters. To tackle this problem, we present a new technique for 3D geometry reconstruction by using inverse synthetic aperture radar (ISAR) image sequence energy accumulation (ISEA). Firstly, by constituting the motion model of SRST, an explicit expression is derived to describe the relative geometric relationship between the 3D geometry and ISAR image sequence. Then accurate rotational parameters and the 3D geometry of SRST can be estimated by combining the idea of the ISEA method and quantum-behaved particle swarm optimization (QPSO). Compared with the ISEA method, which can be only applied to triaxial stabilized space targets, the proposed method can achieve 3D geometry reconstruction of SRST. Experimental results based on the simulated point model and simulated electromagnetic computer aided design (CAD) model validate the effectiveness and robustness of the proposed method.

Online Wear Particle Detection Sensors for Wear Monitoring of Mechanical Equipment—A Review

Online wear particle detectors have been widely studied due to their ability to monitor the wear state of machines, and to help to prevent serious mechanical failures caused by excessive wear of components. This paper presents a review of the state-of-the-art wear debris detectors based on different principles. That mainly includes optical/imaging particle detectors, electrical particle detectors, ultrasonic particle detectors and magnetic debris detection sensors. Meanwhile, the characteristics and the performance (sensitivity, maximum flow rate, and the detectable information) of each type of sensor are detailed discussed. In conclusion, the optical/imaging and magnetic debris detectors have become promising technologies in the field of wear monitoring. Meanwhile, that the rapid extraction algorithm and three-dimensional reconstruction method of wear debris for optical/imaging debris detectors, and improving the sensitivity and detectability of magnetic debris detectors are the two important research contents in the field. Besides that, it can be predicted that the integrated wear monitoring system that can collect multiple wear-information in-situ, and the intelligent wear particle detector that have the ability of wear monitoring, wear evaluation and fault warning are the main research directions in the future.

Deep learning based automated segmentation of air-void system in hardened concrete surface using three dimensional reconstructed images

The automated air-void detection methods specified in the ASTM C457 require the aid of contrast enhancement which is time consuming and labor intensive. This study investigated the utilization of three-dimensional (3D) reconstruction and Deep Convolution Neural Network (DCNN) methods to detect the air voids in hardened concrete surfaces without the use of contrast enhancement. The experimental results showed that the DCNN could accurately distinguish air voids from hardened concrete images with the detection accuracy of over 0.9 in only less than a minute. The accuracy rates for air content, specific surface, and spacing factor were 0.92, 0.91, and 0.89, respectively.

Learning Pyramidal Feature Hierarchy for 3D Reconstruction

Neural network-based three-dimensional (3D) reconstruction methods have produced promising results. However, they do not pay particular attention to reconstructing detailed parts of objects. This occurs because the network is not designed to capture the fine details of objects. In this paper, we propose a network designed to capture both the coarse and fine details of objects to improve the reconstruction of the fine parts of objects.

Convergence of Stereo Vision-Based Multimodal YOLOs for Faster Detection爋f Potholes

Road potholes can cause serious social issues, such as unexpected damages to vehicles and traffic accidents. For efficient road management, technologies that quickly find potholes are required, and thus researches on such technologies have been conducted actively. The three-dimensional (3D) reconstruction method has relatively high accuracy and can be used in practice but it has limited application owing to its long data processing time and high sensor maintenance cost. The two-dimensional (2D) vision method has the advantage of inexpensive and easy application of sensor. Recently, although the 2D vision method using the convolutional neural network (CNN) has shown improved pothole detection performance and adaptability, large amount of data is required to sufficiently train the CNN. Therefore, we propose a method to improve the learning performance of CNN-based object detection model by artificially generating synthetic data similar to a pothole and enhancing the learning data. Additionally, to make the defective areas appear more contrasting, the transformed disparity map (TDM) was calculated using stereo-vision cameras, and the detection performance of the model was further improved through the late fusion with RGB (Red, Green, Blue) images. Consequently, through the convergence of multimodal You Only Look Once (YOLO) frameworks trained by RGB images and TDMs respectively, the detection performance was enhanced by 10.7% compared with that when using only RGB. Further, the superiority of the proposed method was confirmed by showing that the data processing speed was two times faster than the existing 3D reconstruction method.

Research on Multi-View Image Reconstruction Technology Based on Auto-Encoding Learning

Traditional three-dimensional (3D) image reconstruction method, which highly dependent on the environment and has poor reconstruction effect, is easy to lead to mismatch and poor real-time performance. The accuracy of feature extraction from multiple images affects the reliability and real-time performance of 3D reconstruction technology. To solve the problem, a multi-view image 3D reconstruction algorithm based on self-encoding convolutional neural network is proposed in this paper. The algorithm first extracts the feature information of multiple two-dimensional (2D) images based on scale and rotation invariance parameters of Scale-invariant feature transform (SIFT) operator. Secondly, self-encoding learning neural network is introduced into the feature refinement process to take full advantage of its feature extraction ability. Then, Fish-Net is used to replace the U-Net structure inside the self-encoding network to improve gradient propagation between U-Net structures, and Generative Adversarial Networks (GAN) loss function is used to replace mean square error (MSE) to better express image features, discarding useless features to obtain effective image features. Finally, an incremental structure from motion (SFM) algorithm is performed to calculate rotation matrix and translation vector of the camera, and the feature points are triangulated to obtain a sparse spatial point cloud, and meshlab software is used to display the results. Simulation experiments show that compared with the traditional method, the image feature extraction method proposed in this paper can significantly improve the rendering effect of 3D point cloud, with an accuracy rate of 92.5% and a reconstruction complete rate of 83.6%.

Three-Dimensional Reconstruction Method Based on Speckle and Phase Hybrid Light Field Modulation

Three-Dimensional Reconstruction Method Based on Speckle and Phase Hybrid Light Field Modulation

APPLICATION OF AERIAL PHOTOGRAPHY RESULTS, OBTAINED USING UAV DATA TO ESTIMATE THE RESIDUAL CAPACITY OF THE LANDFILL. PRACTICE OF THE REPUBLIC OF TATARSTAN

Assessment of the current residual capacity is a fundamentally important task, the solution of which is demonstrated on a landfill located in the Republic of Tatarstan. To solve the task, the modern methods of high-precision three-dimensional reconstruction were used based on the survey from an unmanned aircraft DJI Phantom 4, equipped with a global satellite navigation system (GNSS) receiver. As a result of combining the project data and data from field surveys into one coordinate system and elevations and reconstruction of designed underground and ground parts of landfill calculation, the difference of models and the residual capacity of the landfill becomes possible. Based on the materials considered as of July 2020, the residual capacity of the studied landfill is 41.2 % of the project capacity, which allows us to continue to operate this landfill. The proposed approach allows for rapid and high-quality active monitoring of the engineered facility. Photogrammetric processing of the results of low-altitude aerial photography makes it possible to obtain objective data on the current actual state of the landfills, to carry out competent and valid management of the landfill functioning. Significant, this will extend the landfill's lifetime, minimize the adverse effects on the environment and predict the yield to the project capacity much more accurately.

Quantitative anatomy of primary ossification centres of the lateral and basilar parts of the occipital bone in the human foetus.

Computed tomography (CT)-based quantitative analysis of primary ossification centres in the cranium has not been carried out to date due to the limited availability of the foetal human material. Detailed morphometric data about the development of primary ossification centres in human foetuses may be useful in the early detection of developmental defects. Understanding the growth and development of the occipital bone is crucial in assessing the normal and pathological development of the cranial base, and the cranium as a whole. The study material comprised 37 human foetuses (16 males and 21 females) aged 18-30 weeks of gestation. Using CT, digital image analysis software, three-dimensional reconstruction and statistical methods, the size of the primary ossification centres of the lateral and basilar parts of the occipital bone was evaluated. The morphometric characteristics of primary ossification centres of the lateral and basilar parts of the occipital bone display neither sex nor laterality differences. These ossification centres grow linearly with respect to their sagittal and transverse diameters, projection surface area and volume. The obtained morphometric data of primary ossification centres in the lateral and basilar parts of the occipital bone may be considered as normative for their respective prenatal weeks and may contribute to the estimation of gestational ages and the diagnostics of congenital defects.

Effects of Different Parameter Settings for 3D Data Smoothing and Mesh Simplification on Near Real-Time 3D Reconstruction of High Resolution Bioceramic Bone Void Filling Medical Images.

Three-dimensional reconstruction plays a vital role in assisting doctors and surgeons in diagnosing the healing progress of bone defects. Common three-dimensional reconstruction methods include surface and volume rendering. As the focus is on the shape of the bone, this study omits the volume rendering methods. Many improvements have been made to surface rendering methods like Marching Cubes and Marching Tetrahedra, but not many on working towards real-time or near real-time surface rendering for large medical images and studying the effects of different parameter settings for the improvements. Hence, this study attempts near real-time surface rendering for large medical images. Different parameter values are experimented on to study their effect on reconstruction accuracy, reconstruction and rendering time, and the number of vertices and faces. The proposed improvement involving three-dimensional data smoothing with convolution kernel Gaussian size 5 and mesh simplification reduction factor of 0.1 is the best parameter value combination for achieving a good balance between high reconstruction accuracy, low total execution time, and a low number of vertices and faces. It has successfully increased reconstruction accuracy by 0.0235%, decreased the total execution time by 69.81%, and decreased the number of vertices and faces by 86.57% and 86.61%, respectively.

Comparison of Different Tissue Clearing Methods for Three-Dimensional Reconstruction of Human Brain Cellular Anatomy Using Advanced Imaging Techniques.

The combination of tissue clearing techniques with advanced optical microscopy facilitates the achievement of three-dimensional (3D) reconstruction of macroscopic specimens at high resolution. Whole mouse organs or even bodies have been analyzed, while the reconstruction of the human nervous system remains a challenge. Although several tissue protocols have been proposed, the high autofluorescence and variable post-mortem conditions of human specimens negatively affect the quality of the images in terms of achievable transparency and staining contrast. Moreover, homogeneous staining of high-density epitopes, such as neuronal nuclear antigen (NeuN), creates an additional challenge. Here, we evaluated different tissue transformation approaches to find the best solution to uniformly clear and label all neurons in the human cerebral cortex using anti-NeuN antibodies in combination with confocal and light-sheet fluorescence microscopy (LSFM). Finally, we performed mesoscopic high-resolution 3D reconstruction of the successfully clarified and stained samples with LSFM.

Features of the national development of possibilities forensic diagnostics of the gunshot injuries

This article shows the role and contribution of individual forensic scientists and famous forensic academies in the development of learning gunshot injuries. The article presents an analysis of the national literary sources concerning development of possibilities and methods of forensic diagnostics of the main traumatic factor of gunshot injuries. The role of up-to-date digital technologies, particularly the method of three-dimensional spatial reconstruction of the gunshot wounds, which can empower possibilities of forensic diagnostics of the main traumatic factor of gunshot injuries are shown.
 Aim of the work. To analyze the development of diagnostic methods of a kind of the main traumatic factor of gunshot injuries in forensic medicine, to find the up-to-date and most diagnostically valuable methods for further investigation and implementation into experts’ practical work.
 Conclusions. Analysis of national forensic experts’ studies shows that forensic medical examination of gunshot body wounds has always been of current interest and become more and more actual. In every period of forensic medicine development new instrumental and laboratory methods of examination of gunshot wounds have appeared. One of the most topical and promising methods are means of the three-dimensional spatial reconstruction. It can improve the possibilities of forensic medical diagnostics of a kind and characteristic of the main traumatic factor of gunshot wounds.

The three-dimensional reconstruction of the dilated renal pelvicalyceal system by non-enhanced computed tomography

Introduction. The three-dimensional reconstruction of the renal pelvicalyceal system (PCS) is possible when performing enhanced computed tomography (CT). However, the use of a contrast agent has its limitations associated with the presence of allergy and chronic kidney disease.Purpose of the study. To describe the method of semi-autonomous three-dimensional (3D) reconstruction of the PCS based on non-enhanced CT images of patients with upper urinary tract obstruction.Materials and methods. Five patients diagnosed with renal colic were recruited from April-May 2021. All patients underwent CT-urography after informed consent. Medical Imaging Interaction Toolkit program (MITK) expanded with explainable update were used for 3D-reconstruction of PCS via excretory and native phases. To assess the accuracy of the latter, both contrast and non-contrast models were compared regarding their surface area. Also, the PCS of one patient was used to reconstruct virtual endoscopic views based on enhanced and non-enhanced models. Five urologists estimated their similarity and potential use of non-enhanced models for the interventional planning via a Likert scale questionnaire. The resulting models were also analyzed by programmer-engineers to test their suitability for 3D-printing.Results. The average surface area of enhanced and non-enhanced models was 3291 mm2 and 2879 mm2, respectively. Obtained models were suitable for their intraluminal reconstruction and potential 3D-printing. Analyzed properties of non-enhanced models were estimated at 4.5 out of 5.0.Conclusion. The described semi-autonomous reconstruction of the renal PCS based on non-enhanced CT images allows for a short time to reconstruct its 3D-view in patients with the upper urinary tract obstruction.