Three-Dimensional Scene Reconstruction: A Review of Approaches

Three-dimensional scene reconstruction is an essential task in computer vision and robotics, involving methods to create a three-dimensional representation of a scene from two-dimensional images. Various approaches have been developed, including shape from silhouette, stereo vision, and fusion techniques. This review outlines the main methods, their advantages and limitations, and provides references for further research.

Developing Concepts and Applications

The perception of 3D scene with stereovision is the capability of human vision but it is a challenge to computer systems. The challenge is to obtain 3D geometric shape information from planar images. This is termed as 3D reconstruction. It is usually done in a piece-wise fashion by identifying various planes of the scene and from those planes constructing a representation of the whole. The image is captured from a calibrated camera. The captured image is perspectively distorted. These distortions are removed by corner point estimation. The selected points give the true dimensions in the input image using view metrology. Then a 3D geometrical model is constructed in VRML according to true dimensions from metrology. In rendering, a texture map is created to each corresponding surface of a polygon in VRML model. VRML supports for walkthrough to a rendered model, through which different views are generated.

Journal of Information Processing Systems, 2016

The recent advent of increasingly affordable and powerful 3D scanning devices capable of capturing high resolution range data about real-world objects and environments has fueled research into effective 3D surface reconstruction techniques for rendering the raw point cloud data produced by many of these devices into a form that would make it usable in a variety of application domains. This paper, therefore, provides an overview of the existing literature on surface reconstruction from 3D point clouds. It explains some of the basic surface reconstruction concepts, describes the various factors used to evaluate surface reconstruction methods, highlights some commonly encountered issues in dealing with the raw 3D point cloud data and delineates the tradeoffs between data resolution/accuracy and processing speed. It also categorizes the various techniques for this task and briefly analyzes their empirical evaluation results demarcating their advantages and disadvantages. The paper concludes with a cross-comparison of methods which have been evaluated on the same benchmark data sets along with a discussion of the overall trends reported in the literature. The objective is to provide an overview of the state of the art on surface reconstruction from point cloud data in order to facilitate and inspire further research in this area.

The task of generating fast and accurate 3D models of a scene has applications in various computer vision fields, including robotics, virtual and augmented reality, and entertainment. So far, the computer vision scientific community has provided innovative reconstruction algorithms that exploit variant types of equipment. In this paper a survey of the recent methods that are able to generate complete and dense 3D models is given. The algorithms are classified into categories according to the equipment used to acquire the processed data.

2015

The increase in demand for 3D content has inspired researchers to devise techniques which allow 3D models to be acquired directly from the images. Various techniques have been developed to generate 3D model from 2D images. This paper covers various approaches for 3d generation. Model based techniques use image features such as texture, shading for generating 3D model. Constraint-based techniques use geometric properties like co-linearity, coplanar, point, normal. The geometric approach deals with geometric relationships between points, lines, planes, etc under imaging. Keywords—3D Reconsruction, Shape from silhouette, 3D mesh

2011

The Journal of Engineering Research, 2017

3D reconstruction from multiple views is well studied, fundamental, yet a challenging problem in the field of computer vision. There is a large variety of approaches available in the literature. The methods use different representations for input scene/object and may provide different kinds of outputs. Some methods model entire scene as voxel-set where as some use level sets or polygon mesh representation. Output may be either volume or surface representing the reconstructed object/scene. Some methods work in image space where as some methods work in object space. These methods are developed to offer a good compromise between computation speed, computation complexity and accuracy along with feasibility in implementation. Selection of a particular method depends on the requirements of application and availability of required resources. Though, earlier reviews are available in the literature, fast advances in this field demand latest review. The paper presents a review and comparison ...

2017 8th International Conference on Computing, Communication and Networking Technologies (ICCCNT), 2017

This paper presents an efficient method to reconstruct 3D point cloud from an arbitrary number of views of an object of interest. Nowadays, Light Detection and Ranging (LIDAR) devices are used in autonomous vehicles and robots to get the 3D measurements. LIDAR contains laser scanners which directly gives the 3D information. However, these scanners are very expensive and prone to interferences and noise due to bulk hardware present in the device. To address this problem, mobile phone cameras are used in the proposed method as they are lighter and cheaper compared to scanners. Multiviews of an object of interest are obtained using a mobile phone camera and they are given as input to the structure from motion (SFM) algorithm. In this algorithm, an appropriate set of mixed feature extraction techniques are employed to get good number of 3D inlier points. The obtained results are visualised as 3D point clouds.

In this paper, we propose a methodology for 3D virtual reconstruction of objects that can be applied to virtual restoration. The methodology is based on an image-based modelling technique and allows generating a textured 3D mesh from a set of images. The proposed methodology consists in the following actions: obtain images of the object, processing of the images, 3D reconstruction of the object, finishing and completing details, restoration of the 3D virtual model. First, we review several frameworks and toolkits that can be used for image-based modelling and then a detail example of 3D reconstruction is presented. The advantage of this methodology for 3D virtual reconstruction is the use of inexpensive equipment, because only common video and computing devices are needed.

2018

Estimation of 3-D objects from 2-D images is inherently performed by either motion or scene features methods as it has been described in different literatures. Structure from motion as a method employed in this study uses calibrated camera and reconstructed 3-D points from the structure of the scene for reliable and precise estimates. In this study we construct 3-D shapes using color pixels and reconstructed 3-D points to determine observable differences for the constructed 3-D images. The estimation using reconstructed 3-D points indicates that the sphere is recovered by the use of scale factor due to its known size while the one obtained by using color pixels has look similar to the former but different in the scales of the axes.

AI

International Journal of Electrical and Computer Engineering (IJECE), 2019

3D reconstruction are used in many fields starts from the object reconstruction such as site, cultural artifacts in both ground and under the sea levels, medical imaging data, nuclear substantional. The scientist are beneficial for these task in order to learn, keep and better visual enhancement into 3D data. In this paper we differentiate the algorithm used depends on the input image: single still image, RGB-Depth image, multiperspective of 2D images, and video sequences. The prior works also explained how the 3D reconstruction perform in many fields and using various algorithms. 1. INTRODUCTION 3D Reconstruction task is one of the interesting task that meet its maturity already. These can be seen from the commercial products such as product from Agisoft and Pix4D that are capable of produced high quality of large scale 3D models. Furthermore, the hardware such as the computer vision has been developed and improve since then. There are some setup camera introduced in the research such as stereo camera and Kinect. In addition to the vision setup, kinect camera shows a great positive feedback from the researchers, proved by common vision setup that can be found in the literature review. Not only that, stereo camera setup can be found among the literature review. In addition to the stereo camera, custom stereo camera are quite popular among the researchers by combining two equals web camera that positioned by period of distance. The algorithm to perform 3D reconstruction between these camera are different due to the produced images are different as well. Kinect abilities allows RGB image and depth map produced, on the other hand Stereo camera has to perform another depth map acquisition algorithm by combining 2 RGB images. Numerous numbers of 3D reconstruction task can be found in capturing the site, cultural artifacts both in ground and under the sea levels [1]. The extinction factor is the most prominent issue in these area. Moreover, 3D imaging data also could help improve the accuracy of the anatomical features in order to observe some areas before coming to the surgery action .Furthermore, in order to perform 3D reconstruction, there are multiple approaches found in the literature review such as from the broad ranges of vision setup, various types of inputted image to construct 3D reconstruction. Thus, In this paper will describe more on those approaches. The great numbers of the researchers along with the hardware supports allows such algorithm to do high processing calculation in order to perform reconstruction task. There are some sections mentioned in part 2.. The benefits of reconstruction are to perform 3D recording, visualization, representation and reconstruction [2]. Moreover Tsiafaki and Michailidou explained that, there are 6 benefits in performing reconstruction and visualization: limiting the destructive nature of excavating, placing excavation data into the bigger picture, limiting fragmentation of archaeological remains, classifying archaeological finds, limiting subjectivity and publication delays, enriching and extending archaeological research.

Advances in multimedia and interactive technologies book series, 2018

Over the past years, 3D reconstruction has proved to be a challenge. With augmented reality and robotics attracting more attention, the demand for efficient 3D reconstruction algorithms has increased. 3D reconstruction presents a problem in computer vision and as a result, much work has been dedicated to solving it. Different design choices were made to consider different components of the process. Examples of these differences are how the scanning process is tackled, how the 3D reconstructed world is represented, among other aspects. Therefore, an evaluation of these algorithms is necessary. This chapter focuses on the properties that facilitate the evaluation of 3D reconstruction algorithms and provides an evaluation of the various algorithms.

– The extract key points and matching the pictures are the most paramount reconstruction 3D factors. They almost two-thirds the time of reconstruction. This paper presents a method to extract the most paramount key points, through the use of GrabCut algorithm that elimintes considerable parts of images that does not have its prominence in the reconstructio. Moreover, the proposed algorithm uses siftGPU algorithm that runs parallel to any process more than one image at a time to extract key points and carry out matching process. The experiments show that the proposed system increase the speed of reconstruction and thoroughly good. Keywords – 3D Reconstruction, S tructure From Motion (S FM), Mash Reconstructionand Multi-View S tereo (MVS). I. INTRODUCTİON 3D reconstruction is one of the classical and difficu lt problems in co mputer vision, and finds its applications in a variety of different fields. In recent years, large scale 3D reconstruction from co mmunity photo collections has become an emerging research topic, wh ich is attracting more and more researchers fro m academy and industry. However, 3D reconstruction is extremely co mputationally expensive. For examp le, it may cost more than a day in a single mach ine to reconstruct an object with only one thousand pictures. In the Structure from Motion (SfM) model [1, 2], 3D reconstruction pipeline can be divided into various steps: feature extraction, image matching, track generation and geometric estimation, etc. Among them, image matching occupies the fundamental computational cost, even more than half of all in some case. Moreover, inexact matching results might lead to washout of reconstruction. Therefore, fast and accurate image matching is crit ical for 3D reconstruction. There are various ways to build reconstruction For example Reconstruction manually is most Statute method to reconstruct a 3D model for an object real world. but is a method ponderous and very intensive. Level of realism can be achieved [3]. The other way tried to eliminate the voltage on the user. 3D Scanner Variant Gu ide to reconstruction is to let co mputers to take some work, and is a well-established method of 3D scanning. The 3D the scanner is the device that apprehends the detailed informat ion for shape and appearance[4]. Modern developments in techniques scanners and Laser able to apprehend point clouds of scenes the real world, And also Automatically can reveal scene planes and create 3D models without the help of the user which can generate Points dense cloud from total images by photogrammetry tools [5].To create a point clouds typically sharing the same problems fro m noisy and lost data. Makes it very hard to apply the methods of surface reconstruction the direct [6,7], Points cloud doesn't contain the specific edges and borders. Last method offered by the this paperPhotogrammetry reconstruction regains 3D informat ion fro m a single or more of images. Main ly focused on rebuilding Photos mu lti view called stereo vision. Epipolar geo metry describes the features and the linkages between the scene and the 3D geometric projections on two or more images of 2D. Figure 1 shows the idealistic workflow for photogrammetric reconstruction. The first step of photogrammetric reconstruction includes the registration of all input images. This procedure is called structure-fro m-mot ion and includes the computation of intrinsic and extrinsic camera parameters. For reg istered images, it is possible to compute 3D positions from t wo or mo re corresponding image points. Multi-view stereo algorithms use these conditions and compute dense point clouds or triangulated meshes fro m the input scene. A B Fig.1. Reconstruction Photogrammet ry are recording mu ltip le images (A): is created by the structure fro m motion (B) and 3D geo metry by dense mult i view stereo [8]. The terms Multi-view Stereo (MVS) simu lates the sense of human sight distance and 3D objects. It uses two or more images fro m various points of view to get the 3D structure of the scene and distance information. Many algorith ms stereo mult iview [9, 10] used all the images at the same time to rebuild the 3D model. It requires a high expense and also lacks scalability. Furukawa [ 10] suggested PMVS (mult iple stereoscopic vision correction) and took mu ltiple p ictures of various views of the body to extract feature points. It is then expanded abroad to find more of the interview points. Furukawa CM VS also suggested (Views comp ilation of mult iple stereo) [ 12] in 2010, has been used to ameliorate the image co mb ines numerous susceptibility in order to see the stereo, and sustainable forest management PM VS b roker contacts. RGB-D systems have been developed due to the advent of RGB-D sensors, such as the Microsoft Kinect.

3D RECONSTRUCTION, 2020

Multi software used to get the results of multiple images. The (SFM) and (CMVS) used to obtain the sparse and dense reconstruction of sequential images collections. By taken images of engine through robot from different view to quantify the quality of (SFM) data in relation. We have implemented a number of comparison b/w different parts of engine in order to assess the mesh deviation and the reconstruction accuracy. The (SFM) and (CMVS) techniques and methods were mostly questioned by creating a complete 3D digital replication of engine.

International Journal for Research in Applied Science & Engineering Technology (IJRASET), 2023

Using deep neural networks, supervised 3D reconstruction has made significant progress. However, large-scale annotations of 2D/3D data are necessary for this performance boost [4]. How to effectively represent 3D data to feed deep networks remains a challenge in 3D deep learning. Volumetric or point cloud representations have been used in recent works, but these methods have a number of drawbacks, including computational complexity, unorganized data, and a lack of finer geometry [7]. The ultrasound (US) examination is one method used to diagnose carotid artery disease. The flow conditions in the artery also play a role in the onset and progression of vascular diseases [5]. Stenting the carotid artery is a treatment option for carotid atherosclerosis. It is impossible to know for sure where an injection with a needle will begin. The place of the conduits is in the body, thusly, deciding the beginning stage of needle infusion is finished by assessment just and can't not entirely set in stone. The first thing that must be done is to locate the carotid artery in order to identify it. To determine it, we propose a modified template matching based on the ellipse feature for a 3D reconstruction of the carotid artery [1]. Data acquisition, pre-processing, segmentation, outlier selection for ellipse parameter fitting, and visualization are all used to process it. In comparison to the template matching method and the Hough Circle method, the proposed procedure with pre-processing produces the highest accuracy [1]. The objective of this research was to create three-dimensional (3D) ultrasound imaging of the carotid arteries to lessen the variability of volume measurements between and within examiners during follow-up scans of atherosclerotic plaques.

International Journal for Research in Applied Science and Engineering Technology, 2023

The review paper emphasize on reducing technologies for 2D and 3D imaging, as well as model conversion. Although the popularity of 3D hardware is growing rapidly in the present era, 3D content is still dominated by its 2D counterpart. There are two main categories of image processing now available in the market, namely analogue and digital image processing. To produce hard copies such as scanned pictures and printouts, with images being the most common output, the analogue IP technique is used. On the other hand, Digital IP is used to manipulate digital images using computers, the outputs are often information related to images, mainly being data on features, edging characteristics, or masks. Image processing techniques, including Machine Learning and Deep Learning, can get more powerful.

2011

This paper presents a comprehensive study of all the challenges associated with the design of a platform for the reconstruction of 3-D objects from stereo images. The overview covers: Design and Calibration of a physical setup, Rectification and disparity computation for stereo pairs, and construction and rendering of meshes in 3-D for visualization. We outline some challenges and research in each of these areas, and present some results using some basic algorithm and a system build with off-the-shelf components.

ABCM Symposium Series in Mechatronics, 2006

The necessity of obtaining geometric models in three-dimension that represent with precision a real world object is becoming common each day. For this, one has to recur to methods of 3D Modeling. Three-dimension models have application in several areas, amongst which one can cite photogrammetry, archaeology, reverse engineering, robotic guidance, virtual reality, medicine, cinema, game programming, and others. A current challenge is the construction of 3D models digitized with precision enough to be used in manufacturing systems or numerical simulation of the performance of machines and components in operation, such as turbines and flows in non-circular ducts when the geometric model is not available. The reconstruction of 3D shapes of objects or scenes from range images, also known as depth maps, is preferable than using intensity images or stereoscopy. These maps represent information of distances measured from an observer (optical sensor or camera) to the scene in a rectangular grid. Therefore, the 3D information is explicit and will not need to be recovered as in the case of intensity images. The reconstruction process presents three stages. The first one is sampling of the real world in depth maps. The second stage is the alignment of several views within the same coordinate system, known as image registration. The third stage is the integration of the views for the generation of surface meshes, named merging. The current challenges converge to searching methods that meet with the highest number of desirable properties, such as robustness to outliers, efficiency of time and space complexity and precision of results. This work consists in the discussion of different methods dealing with 3D shape reconstruction from range images found in the literature and in the implementation of the second phase of 3D reconstruction: range image registration.

2012

Digital images provide today an important source of data that deserves a careful statistical analysis. This paper concerns methods for retrieval of 3D information, including shape and texture, from cheap digital camera imaging outputs. It includes a three step reconstruction of a 3D scene with texture, from arbitrary partial views, in absence of occlusions. In Patrangenaru and Patrangenaru [23], Mardia et. al. and Patrangenaru and Mardia [24] a planar scene was reconstructed using image fusion, around representatives of sample mean projective shapes or sample mean affine shapes of landmark configurations shared by a number of partial views of the scene. In this paper we first analyze the advantages and limitations of such a reconstruction of a close to planar remote scene from its partial aerial views, by specializing this algorithm to affine transformations. Furthermore, we combine a projective shape reconstruction of a finite 3D configuration from its uncalibrated camera views, as developed in Patrangenaru, Liu and Sughatadasa [22], with a VRML technique, to reconstruct projectively a 3D scene with texture from a pair of digital camera images, thus allowing a more detailed statistical analysis of the scene pictured. We give three such examples of 3D reconstructions.

• Explore
• Papers
• Topics

• Features
• Mentions
• Analytics
• PDF Packages
• Advanced Search
• Search Alerts

• Journals
• Academia.edu Journals
• My submissions
• Reviewer Hub
• Why publish with us
• Testimonials

• Company
• About
• Careers
• Press
• Help Center
• Terms
• Privacy
• Copyright
• Content Policy