21 January 2012 in The International Journal of Advanced Manufacturing Technology
25 April 2019 in IEEE Transactions on Multimedia
3D pose estimation for texture-less objects remains a challenging problem. Previous works either focus on a template matching method to find the nearest template as a candidate, or construct a Hough Forest, which utilizes the offset of patches to vote for the object location and pose. By contrast, in this paper, we propose a comprehensive framework to directly regress 3D poses for the candidates, in which a CNN-based triplet network is trained to extract discriminated features from the binary images. To make the features suitable for the regression task, a pose-guided method and a regression constraint are employed to the constructed triplet network. We show that the constraint reaches the goal of creating the correlation between the features and 3D poses. Once the expected features are obtained, the object pose could be efficiently regressed by training a regression network with a simple structure. For symmetric objects, depth images are treated as an additional channel to feed the triplet network. Experiments on the LineMOD and our own datasets demonstrate our method with high regression precision and efficiency.
10 January 2013 in Journal of Computing and Information Science in Engineering
24 June 2022 in IEEE Transactions on Instrumentation and Measurement
The pipe is a common part of aviation equipment. The pipe gap is one of the important factors that affect the safety of an aviation equipment system. It is crucial to quickly and accurately measure the gaps among the pipes within the aviation equipment. Due to the complexity of the aviation pipeline system, inspecting the spacing efficiently among different pipes is challenging. In this article, we propose an effective method to inspect the gaps among aircraft pipes over the raw point data from a single view. First, we design a probe function to extract the information of pipe structures based on the cylinder feature consistency among proximity patches. The probe function is intended to avoid the interference of other structures and also provides an appropriate starting location when extracting the pipe structure information. We then design a growth guidance strategy based on the continuity of pipe axial directions, which can ensure the continuity and accuracy of pipe centerline extraction. Since the curved smooth pipe can be subdivided into many straight cylinders, we devise an adaptive growing method, which can ensure the centerline extraction efficiency of multicurvature aviation pipes. Extensive experiments demonstrate that our algorithm outperforms the state-of-the-art methods in terms of inspection accuracy and efficiency.
4 March 2024 in IEEE Transactions on Instrumentation and Measurement
To enable the generation of reliable tool paths for precision machining along the edges of thin-walled aircraft assembly panels, we introduce a feature-aware displacement learning framework for accurately extracting the edges of aircraft panels. Specifically, we design a dual-task neural network, named APEE-Net. This network serves the dual purpose of identifying points located near the edges of aircraft panels and predicting displacement vectors pointing toward local edge features. The detected edge points are subsequently re-positioned using these displacement vectors, resulting in the extraction of precise edge points. Our proposed method is fortified with feature-aware displacement optimization loss during the training phase, significantly enhancing its robustness and accuracy when dealing with noisy sharp geometric features. Extensive experiments demonstrate that our approach outperforms existing extraction methods in terms of accuracy. Furthermore, practical machine applications further validate its feasibility and real-world applicability.
30 June 2021 in IEEE Robotics and Automation Letters
Scene reconstruction with consumer-level RGB-D cameras has developed considerable momentum in both robotics and vision communities. In the literature of robotics, high-quality camera tracking, the key to accurate reconstruction, is challenging in geometric featureless scenes or under large lighting variation. In this letter, we propose an accurate RGB-D reconstruction approach, by incorporating 3D line features into key steps of the pipeline to significantly improve tracking robustness and reduce error accumulation. Our reconstruction approach is characterized by three key designs. First, we propose a robust 3D line extractor to efficiently extract accurate and consistent 3D lines from RGB-D data. Moreover, we design a joint objective for frame-to-model scan alignment considering both dense points and 3D lines in a robust error metric. With extra constraints from 3D lines and the robust error metric, scan alignment is further enhanced in challenging situations even without any heuristics for spurious correspondence pruning. Finally, we propose an efficient submap-based hierarchical optimization considering both point and 3D line correspondences to significantly reduce the accumulated error in camera tracking and obtain an accurate global 3D model. Extensive experiments on synthetic and real-world datasets demonstrate that our method outperforms state-of-the-art RGB-D reconstruction approaches for common indoor or outdoor scenes in challenging situations of practical scanning.