Search results

The purpose of this paper is to present a novel framework of reconstructing the 3D textile model with synthesized texture.

First, a pipeline of 3D textile reconstruction based on KinectFusion is proposed to obtain a better 3D model. Second, “DeepTextures” method is applied to generate new textures for various three-dimensional textile models.

Experimental results show that the proposed method can conveniently reconstruct a three-dimensional textile model with synthesized texture.

A novel pipeline is designed to obtain 3D high-quality textile models based on KinectFusion. The accuracy and robustness of KinectFusion are improved via a turntable. To the best of the authors’ knowledge, this is the first paper to explore the synthesized textile texture for the 3D textile model. This is not only simply mapping the texture onto the 3D model, but also exploring the application of artificial intelligence in the field of textile.

This paper aims to quickly obtain an accurate and complete dense three-dimensional map of indoor environment with lower cost, which can be directly used in navigation.

This paper proposes an improved ORB-SLAM2 dense map optimization algorithm. This algorithm consists of three parts: ORB feature extraction based on improved FAST-12, feature point extraction with progressive sample consensus (PROSAC) and the dense ORB-SLAM2 algorithm for mapping. Here, the dense ORB-SLAM2 algorithm adds LoopClose optimization thread and dense point cloud map and octree map construction thread. The dense map is computationally expensive and occupies a large amount of memory. Therefore, the proposed method takes higher efficiency, voxel filtering can reduce the memory while ensuring the density of the map and then use the octree format to store the map to further reduce memory.

The improved ORB-SLAM2 algorithm is compared with the original ORB-SLAM2 algorithm, and the experimental results show that the map through improved ORB-SLAM2 can be directly used in navigation process with higher accuracy, shorter tracking time and smaller memory.

The improved ORB-SLAM2 algorithm can obtain a dense environment map, which ensures the integrity of data. The comparisons of FAST-12 and improved FAST-12, RANSAC and PROSAC prove that the improved FAST-12 and PROSAC both make the feature point extraction process faster and more accurate. Voxel filter helps to take small storage memory and low computation cost, and octree map construction on the dense map can be directly used in navigation.

The calibration is a key but cumbersome process for 3D body scanning using multiple depth cameras. The purpose of this paper is to simplify the calibration process by introducing a new method to calibrate the extrinsic parameters of multiple depth cameras simultaneously.

An improved method is introduced to enhance the accuracy based on the virtual checkerboards. Laplace coordinates are employed for a point-to-point adjustment to increase the accuracy of scanned data. A system with eight depth cameras is developed for full-body scanning, and the performance of this system is verified by actual results.

The agreement of measurements between scanned human bodies and the real subjects demonstrates the accuracy of the proposed method. The entire calibration process is automatic.

A complete algorithm for a full human body scanning system is introduced in this paper. This is the first publically study on the refinement and the point-by-point adjustment based on the virtual checkerboards toward the scanning accuracy enhancement.

The aim of this paper is to propose a grasping method based on intelligent perception for implementing a grasp task with human conduct.

First, the authors leverage Kinect to collect the environment information including both image and voice. The target object is located and segmented by gesture recognition and speech analysis and finally grasped through path teaching. To obtain the posture of the human gesture accurately, the authors use the Kalman filtering (KF) algorithm to calibrate the posture use the Gaussian mixture model (GMM) for human motion modeling, and then use Gaussian mixed regression (GMR) to predict human motion posture.

In the point-cloud information, many of which are useless, the authors combined human’s gesture to remove irrelevant objects in the environment as much as possible, which can help to reduce the computation while dividing and recognizing objects; at the same time to reduce the computation, the authors used the sampling algorithm based on the voxel grid.

The authors used the down-sampling algorithm, kd-tree algorithm and viewpoint feature histogram algorithm to remove the impact of unrelated objects and to get a better grasp of the state.

The authors present a low-cost unmanned aerial vehicle (UAV) for autonomous flight and navigation in GPS-denied environments using an off-the-shelf smartphone as its core on-board processing unit. Thereby, the approach is independent from additional ground hardware and the UAV core unit can be easily replaced with more powerful hardware that simplifies setup updates as well as maintenance. The paper aims to discuss these issues.

The UAV is able to map, locate and navigate in an unknown indoor environment fusing vision-based tracking with inertial and attitude measurements. The authors choose an algorithmic approach for mapping and localization that does not require GPS coverage of the target area; therefore autonomous indoor navigation is made possible.

The authors demonstrate the UAVs capabilities of mapping, localization and navigation in an unknown 2D marker environment. The promising results enable future research on 3D self-localization and dense mapping using mobile hardware as the only on-board processing unit.

The proposed autonomous flight processing pipeline robustly tracks and maps planar markers that need to be distributed throughout the tracking volume.

Due to the cost-effective platform and the flexibility of the software architecture, the approach can play an important role in areas with poor infrastructure (e.g. developing countries) to autonomously perform tasks for search and rescue, inspection and measurements.

The authors provide a low-cost off-the-shelf flight platform that only requires a commercially available mobile device as core processing unit for autonomous flight in GPS-denied areas.

Taping, covering objects with masking tapes, is a common process before conducting surface treatments such as plasma spraying and painting. Manual taping is tedious and takes a lot of effort of the workers. This paper aims to introduce an automatic agile robotic system and corresponding algorithm to do the surface taping.

The taping process is a special process which requires correct tape orientation and proper allocation of the masking tape for the coverage. This paper discusses on the design of the novel automatic system consisting of a robot manipulator, a rotating platform, a 3D scanner and a specially designed novel taping end-effectors. Meanwhile, the taping path planning to cover the region of interests is introduced.

Currently, cylindrical and freeform surfaces have been tested. With improvements on new sets of taping tools and more detailed taping method, taping of general surfaces can be conducted using such system in future.

The introduced taping path planning method is a novel method first talking about the mathematical model of the taping process. Such taping solution with the taping tool and the taping methodology can be combined as a very useful and practical taping package to replace the work of human in such tedious and time-consuming works.

The purpose of this paper is to advance a digital technology that is intended to bring about innovations on the existing textile patterns.

The pattern is deemed as a relation function between colors and positions which can be learnt by the artificial neural network (ANN). The outputs of the ANN are used for the reconstruction of the pattern and the innovation is performed by interceptors in the input/output layer. The ANN is carried out with one input layer, one output layer and several hidden layers, and the capacity of the architecture is adjusted by the scale of hidden layers to accommodate different function relations of the patterns. The training is conducted repeatedly on a sample set extracted from the pixels of the pattern image to minimize the error, and the chromatic outputs of the architecture are replaced to their origins so as to rebuild the pattern. Then, the interceptors are installed into the input and output layers to modulate the positions and the colors, and consequently the innovations are achieved on the geometric formation and color distribution of the pattern.

It has turned out that the precision of reconstruction is concerned with network scale, training epochs and color mode of the sample set. Four primary innovative effects including stripes, twisters, sandification and overprints have been qualified in terms of interceptors.

This study introduces ANN into textile pattern generation and provides a novel way to perform digital innovation of textile patterns.