Newsfeeds
Measurement Science and Technology - latest papers
Latest articles for Measurement Science and Technology
-
A sparse representation defect imaging method based on eddy current tomography
As an advanced non-destructive testing technique, eddy current tomography is suitable for defect detection in metallic samples due to the advantages of high sensitivity, fast response speed and intuitive detection results. However, the number of parameters to be solved for in eddy-current tomography is much larger than the number of measured data, which leads to the problem of under characterization of the inversion process. In this paper, a sparse representation reconstruction algorithm based on sigmoid radial basis functions is investigated. Specifically, the basis function vectors of discrete grid points in the imaging region are used to form a basis function matrix. An iterative algorithm is used to solve the fitting coefficients, which can approximate the electromagnetic properties of each point. In turn, the electromagnetic parameter distribution can be reconstructed, enabling the achievement of defect imaging. The method effectively improves the local area fitting accuracy and simplifies the complexity of the imaging model. Experiments are conducted to evaluate the performance of the proposed method using different defect distribution. Compared to the classical reconstruction method, the results indicate that the average relative error and correlation coefficient are improved by 10.8% and 36.2%, respectively.
-
A kernel function based on Gaussian surface construction for fast and high-precision extraction of centers of line-structure light
Line-structure light three-dimensional (3D) measurement is widely applied in the field of non-contact 3D metrology. The curve fitting method is a frequently used algorithm for extracting the center of laser stripes. With respect to the calculation of the sampling direction in the curve fitting method, this paper constructs a kernel function for solving the normal direction of the pixel-level center of the laser stripe, based on the characteristic that the line-structure light follows a Gaussian distribution in the normal direction. By calculating the sampling points within a certain range of the pixel-level center using the kernel function, the first-order and second-order derivatives at the center could be directly obtained, thereby obtaining the normal direction at the center. To enhance the robustness of the sampling data in the normal direction and the real-time processing capability, this paper proposed the calculation of adaptive sampling width and prediction based on the region of interest of the Kalman filtered image respectively. The experimental results demonstrated that the proposed method attains a spherical surface fitting accuracy of 0.0251 mm for the standard target ball. The pre-extraction speed of the stripes is increased by 228% compared to global detection, and the sub-pixel center extraction speed reaches 0.91 milliseconds per frame.
-
YOCO: you only calibrate once for accurate extrinsic parameter in LiDAR-camera systems
In a multi-sensor fusion system composed of cameras and light detection and ranging (LiDAR), precise extrinsic calibration contributes to the system’s long-term stability and accurate perception of the environment. However, methods based on extracting and registering corresponding points still face challenges in terms of automation and precision. This paper proposes a novel fully automatic extrinsic calibration method for LiDAR-camera systems that circumvents the need for corresponding point registration. In our approach, a novel algorithm to extract required LiDAR correspondence point is proposed. This method can effectively filter out irrelevant points by computing the orientation of plane point clouds and extracting points by applying distance- and density-based thresholds. We avoid the need for corresponding point registration by introducing extrinsic parameters between the LiDAR and camera into the projection of extracted points and constructing co-planar constraints. These parameters are then optimized to solve for the extrinsic. We validated our method across multiple sets of LiDAR-camera systems. In synthetic experiments, our method demonstrates superior performance compared to current calibration techniques. Real-world data experiments further confirm the precision and robustness of the proposed algorithm, with average rotation and translation calibration errors between LiDAR and camera of less than 0.05∘ and 0.015 m, respectively. This method enables automatic and accurate extrinsic calibration in a single one step, emphasizing the potential of calibration algorithms beyond using corresponding point registration to enhance the automation and precision of LiDAR-camera system calibration.
-
HSC-YOLO: steel surface defect detection model based on improved YOLOv10n
During the production process of steel, the control of surface quality is crucial to the performance of the final molded product, so it is necessary to detect defects on its surface during the production process. Aiming at the problems of low detection accuracy and insufficient feature extraction and fusion capability in steel surface defect detection, a lightweight and multi-scale feature fusion model HSC-YOLO based on the improved YOLOv10n is proposed. Firstly, the backbone feature extraction network is reset using an improved lightweight network structure based on high performance GPU network (HGNetv2) to reduce the model size. Secondly, the multilevel feature fusion module semantics and detail infusion (SDI) is used instead of the two Concat modules in neck to enhance the semantic and detail information in the image. Finally, an iterative attentional feature fusion (iAFF) mechanism is introduced and combined with cross stage partial bottleneck with 2 convolutions (C2f) to solve the problems that occur when features are fused at different scales, especially the feature fusion problem with inconsistent semantics and scale. Test results on the datasets NEU-DET and GC10-DET show that the mean average precision (mAP) of HSC-YOLO improves by 3.9% and 2.1% over the mAP of YOLOv10n, and the detection speed improves by 46.2% and 43.1%, which provides the best detection results compared to other models.
-
Research on lightweight pipeline defects detection algorithm based on attention mechanism
The safe operation of underground pipelines is critical for production and daily life. To enhance the effectiveness of pipeline intelligent detection and evaluation, this study proposes a novel lightweight pipeline defect detection algorithm based on attention mechanism, LA-YOLO. Using YOLOv10n as the baseline model, channel and spatial attention mechanisms are incorporated into the backbone network to significantly enhance the model’s capability in extracting target features. The lightweight fasternet block module is introduced to construct the C2f-LF module, replacing the original C2f module to simplify the network structure. A lightweight coordinate attention shared parameters detection head is developed, combining attention mechanisms with shared convolutional technology. This innovation markedly reduces the number of parameters while maintaining detection accuracy. Additionally, Wise-IOU is adopted as the loss function instead of Complete-IoU, further improving the model’s precision. To achieve additional model compression, channel pruning is applied to the LA-YOLO architecture, and knowledge distillation is used to recover potential accuracy loss. Experimental results on the USDID demonstrate that the proposed model maintains comparable accuracy and efficiency to the baseline YOLOv10n, and reduces model size, parameters, and floating-point operations by 76.5%, 76.5%, and 57.8%, respectively. The final model size is only 1.2 MB, highlighting its strong potential for real-world deployment in resource-constrained pipeline inspection systems. This work provides a robust and practical solution for efficient and scalable pipeline defect detection.