3D reconstruction of laser projective point with projection invariant generated from five points on 2D target

Closed-loop solution method of active vision reconstruction via a 3D reference and an external camera

Laser-plane-based vision reconstruction is an important approach for 3D profile measurement and optical inspection. This study develops a 3D reconstruction method to measure the object with a flexible method. The active vision reconstruction includes an external camera and a 3D reference with an internal camera and laser projector. The 3D calibration object, the external camera, and the 3D reference constitute a closed loop. The internal camera, the 3D calibration object, and the 3D reference compose the other closed loop. The intrinsic invariants of the 3D reference with an internal camera and laser projector are determined by the two loops above. The benefit of the method is to cancel the position constraint for the active vision system with the cubic LED reference and the monocular external camera. Furthermore, as the external camera captures only the images of the 3D reference, the reconstruction system of the projector and the camera is flexible to observe the occlusion location on the measured object. The methods, including the calibration model and reconstruction model of the measurement system, are verified by experiments to demonstrate the potential applications.

Depth Estimation of a Deformable Object via a Monocular Camera

The depth estimation of the 3D deformable object has become increasingly crucial to various intelligent applications. In this paper, we propose a feature-based approach for accurate depth estimation of a deformable 3D object with a single camera, which reduces the problem of depth estimation to a pose estimation problem. The proposed method needs to reconstruct the target object at the very beginning. With the 3D reconstruction as an a priori model, only one monocular image is required afterwards to estimate the target object’s depth accurately, regardless of pose changes or deformability of the object. Experiments are taken on an NAO robot and a human to evaluate the depth estimation accuracy by the proposed method.

Optimization reconstruction method of object profile using flexible laser plane and bi-planar references

An optimization method to reconstruct the object profile is performed by using a flexible laser plane and bi-planar references. The bi-planar references are considered as flexible benchmarks to realize the transforms among two world coordinate systems on the bi-planar references, the camera coordinate system and the image coordinate system. The laser plane is confirmed by the intersection points between the bi-planar references and laser plane. The 3D camera coordinates of the intersection points between the laser plane and a measured object are initially reconstructed by the image coordinates of the intersection points, the intrinsic parameter matrix and the laser plane. Meanwhile, an optimization function is designed by the parameterized differences of the reconstruction distances with the help of a target with eight markers, and the parameterized reprojection errors of feature points on the bi-planar references. The reconstruction method with the bi-planar references is evaluated by the difference comparisons between true distances and standard distances. The mean of the reconstruction errors of the initial method is 1.01 mm. Moreover, the mean of the reconstruction errors of the optimization method is 0.93 mm. Therefore, the optimization method with the bi-planar references has great application prospects in the profile reconstruction.

Profile reconstruction method adopting parameterized re-projection errors of laser lines generated from bi-cuboid references

A flexible reconstruction method, which is based on the line re-projection errors of the laser plane, is presented for the profile recovery. The bi-cuboid references are designed to cover the large view-field of the camera. The local intrinsic and extrinsic parameter matrices of the camera are initially contributed by the RQ decomposition. Then the balance model is demonstrated to obtain the global parameter matrices in view of the refined projection in the camera coordinate system. The flexible laser plane is solved by the Plücker matrices of the projection laser lines that are generated from the homographies of the cubical references and the global parameter matrices. Furthermore, the laser plane and global parameter matrices are improved by the cost function that is constructed by the re-projection errors of the parameterized laser lines on the references. The reconstruction experiments are performed to verify the validity and the accuracy of the optimization method and the initial method. The impact factors of the measurement distance, the reference distance and the test distance are investigated in the experiments. The average reconstruction errors are 1.14 mm, 1.13 mm, 1.15 mm and 1.17 mm in the four groups of experiments, which shows the good application prospect of the profile reconstructions.