Jibin Zhao

Multi-view measurement of industrial geometric features is susceptible to reduced three-dimensional (3D) reconstruction accuracy due to inconsistent imaging quality, local edge localization errors, and unfavorable observation poses. To address these issues, this paper proposes a hybrid-level weighted multi-view measurement method that evaluates the reliability of cross-view observations from two complementary aspects: local edge quality and global observation geometry. At the local level, a subpixel grayscale model is established along the edge-normal direction to derive a local quality weight. At the global level, an observation-geometry weight is defined based on the angle between the feature-plane normal and the camera optical axis. The two weights are then hierarchically normalized and multiplicatively fused for each set of cross-view correspondences. The resulting fused weights are incorporated into multi-view 3D reconstruction, thereby reducing the propagation of errors arising from low-quality local observations and unfavorable viewpoints to the reconstruction. Experiments show that, compared with equal-weight reconstruction, the proposed method reduces the mean error in center-to-center distance measurement by 41.5%, and the mean errors in square-hole width and height measurement by 22.8% and 25.8%, respectively. These results demonstrate the high precision, improved stability, and strong extensibility of the proposed method for industrial planar geometric feature measurement.

The task of point cloud registration, crucial for for computer vision applications, involves estimating the transformation among point clouds. Accurate correspondence establishment is pivotal for its success. While recent superpoint matching methods have demonstrated considerable potential in point cloud registration, they neglect the spatial consistency among superpoints. In response, we introduce a novel correspondence learning approach that capitalizes on the spatial consistency of superpoints for enhanced coarse matching. This method involves extracting geometric features from point clouds to compute saliency scores, facilitating more accurate transformation estimations. Our comprehensive evaluations across both indoor and outdoor datasets affirm that our approach outperforms current state-of-the-art methods, delivering a better performance.

To improve the performance and reduce the weight of new-generation transparent windshields in aircraft, composite structures are commonly used. The thickness of each layer in a composite windshield has a significant impact on its optical properties and load-bearing capacity. However, existing thickness measurement methods for transparent windshields either cannot meet the requirements in terms of range and efficiency, or cannot measure the thickness of each layer in a transparent windshield. In this article, a large-range spectral-domain low-coherence measurement method for composite transparent windshield thickness is proposed. Specifically, 2-D dispersion technology is used to significantly increase spectral resolution, thereby achieving an improvement in measurement range. Then, a spectral stitching and wavenumber calibration method based on automatic fringe generation is proposed to achieve the conversion from spectral domain to spatial domain. Finally, a fully fiber-coupled co-path spectral-domain low-coherence measurement system for transparent windshields is constructed. Measurement experiment of standard measurement blocks and composite transparent windshields demonstrate that the proposed method has a range of up to 41 mm, a speed of up to 440 Hz, and an accuracy of up to 0.005 mm, thus validating the effectiveness of the proposed method.