A 3D scanning system involving multiple RGB-D cameras has the potential to accelerate the reconstruction of an object and improve the measurement accuracy because it can capture an object in a comprehensive way. However, the extrinsic calibration of a multi-RGB-D-camera system is a fundamental and challenging problem, especially in a limited field of view. In this work, a system with four SR300 cameras on a platform with a size of approximately 0.64 m2 and an extrinsic calibration method assisted by a tower calibration pattern with circular markers in a limited field of view are proposed. The Hough transform algorithm is used to identify the centres of the circular markers in a colour image, and then the 3D coordinates are extracted by employing the alignment relationship between the colour image and the depth image. An improved adaptive cuckoo search (IACS) algorithm and unit quaternion are proposed to optimize the extrinsic parameters. The effectiveness of the IACS is verified by a comparison with the singular value decomposition (SVD), standard cuckoo search (CS) and two CS variant algorithms. In addition, the accuracy of the proposed extrinsic calibration method is verified by reconstructing and measuring a human foot and a cube box. The mean errors from the standard values of the length and width of the reconstructed foot mode are 0.18 cm and 0.13 cm, and the diameter of the yellow sphere are 0.17 cm, respectively. All of the experimental results show that the proposed extrinsic calibration method has a high calibration accuracy and good potential application prospects.

An extrinsic calibration method for multiple RGB-D cameras in a limited field of view

Existing commodity RGB-D sensors have a limited field of view compared with a nearly 360° horizontal field of view of Lidar. This article presents a method to extend the field of view by stitching ...

https://journals.sagepub.com/doi/pdf/10.1177/1729881419851665

Stitching of depth and color images from multiple RGB-D sensors for extended field of view

Three-dimensional (3D) reconstruction methods generate a 3D textured model from the combination of data from several captures. As such, the geometrical transformations between these captures are required. The process of computing or refining these transformations is referred to as alignment. It is often a difficult problem to handle, in particular due to a lack of accuracy in the matching of features. We propose an optimization framework that takes advantage of fiducial markers placed in the scene. Since these markers are robustly detected, the problem of incorrect matching of features is overcome. The proposed procedure is capable of enhancing the 3D models created using consumer level RGB-D hand-held cameras, reducing visual artefacts caused by misalignments. One problem inherent to this solution is that the scene is polluted by the markers. Therefore, a tool was developed to allow their removal from the texture of the scene. Results show that our optimization framework is able to significantly reduce alignment errors between captures, which results in visually appealing reconstructions. Furthermore, the markers used to enhance the alignment are seamlessly removed from the final model texture.

Enhancement of RGB-D Image Alignment Using Fiducial Markers.

Single-pass inline pipeline 3D reconstruction using depth camera array

The creation of a textured 3D mesh from a set of RGD-D images often results in textured meshes that yield unappealing visual artifacts. The main cause is the misalignments between the RGB-D images due to inaccurate camera pose estimations. While there are many works that focus on improving those estimates, the fact is that this is a cumbersome problem, in particular due to the accumulation of pose estimation errors. In this work, we conjecture that camera poses estimation methodologies will always display non-neglectable errors. Hence, the need for more robust texture mapping methodologies, capable of producing quality textures even in considerable camera misalignments scenarios. To this end, we argue that use of the depth data from RGB-D images can be an invaluable help to confer such robustness to the texture mapping process. Results show that the complete texture mapping procedure proposed in this paper is able to significantly improve the quality of the produced textured 3D meshes.

Robust Texture Mapping Using RGB-D Cameras.

This paper presents a novel method to estimate the relative poses between RGB-D cameras with minimal overlapping fields of view. This calibration problem is relevant to applications such as indoor 3D mapping and robot navigation that can benefit from a wider field of view using multiple RGB-D cameras. The proposed approach relies on descriptor-based patterns to provide well-matched 2D keypoints in the case of a minimal overlapping field of view between cameras. Integrating the matched 2D keypoints with corresponding depth values, a set of 3D matched keypoints are constructed to calibrate multiple RGB-D cameras. Experiments validated the accuracy and efficiency of the proposed calibration approach.

/pdf/a-novel-method-for-extrinsic-calibration-of-multiple-rgb-d-p43rq0n10b.pdf

A Novel Method for Extrinsic Calibration of Multiple RGB-D Cameras Using Descriptor-Based Patterns.

This letter presents a novel method to estimate the relative poses between RGB-D cameras with minimal overlapping fields of view in a panoramic RGB-D camera system. This calibration problem is relevant to applications such as indoor 3D mapping and robot navigation that can benefit from a 360$^\circ$ field of view using RGB-D cameras. The proposed approach relies on descriptor-based patterns to provide well-matched 2D keypoints in the case of a minimal overlapping field of view between cameras. Integrating the matched 2D keypoints with corresponding depth values, a set of 3D matched keypoints are constructed to calibrate multiple RGB-D cameras. Experiments validated the accuracy and efficiency of the proposed calibration approach, both superior to those of existing methods (800 ms vs. 5 seconds; rotation error of 0.56 degrees vs. 1.6 degrees; and translation error of 1.80 cm vs. 2.5 cm.

A Novel Method for Extrinsic Calibration of Multiple RGB-D Cameras Using Descriptor-Based Patterns

This letter presents a panoramic 3D vision system that was designed specifically for mapping the indoor environment. The system consists of multiple RGBD sensors that cover a 360◦ field of view, simplifying the reconstruction of indoor environments. A new extrinsic calibration approach is proposed to estimate the relative poses between the RGBD sensors. This approach relies on well-matched keypoints from a descriptorbased pattern, in this case, of a minimally overlapping field of view between sensors. A distributed capturing system was also designed to provide panoramic RGBD frames in real time. Experiments validate the accuracy of the proposed calibration approach and the efficiency of the proposed panoramic 3D vision system for indoor mapping.

Xiahua Liu

Papers

A Novel Method for Extrinsic Calibration of Multiple RGB-D Cameras Using Descriptor-Based Patterns.

A Novel Method for Extrinsic Calibration of Multiple RGB-D Cameras Using Descriptor-Based Patterns

A Panoramic 3D Vision System from Multiple RGBD Sensors.