Efficient and realistic 3D tree modeling is an important part of low-altitude remote sensing. Trees have high geometric complexity, and the reconstruction effect of traditional modeling methods is not satisfactory. The 3D Gaussian Splatting (3DGS) method is expected to achieve good results in 3D reconstruction of trees at a low cost. A new method based on 3DGS, SAGStree, is proposed for 3D tree componentization model, which mainly realizes the segmentation and reconstruction of tree trunk, branch, and leaves. Specifically, to suppress high-frequency artifacts and enhance the expression of foreground features, a semantic-guided smoothing filter is introduced in the 3DGS training process. Meanwhile, to improve the segmentation accuracy of each tree components, the Multi-scale Fourier Attention Aggregation Network (MFAANet) is constructed, which includes a Multiscale Adaptive Fourier module, a Pyramid Attention Aggregation module, and a Dual-Path Decoder module. In addition, a multi-view mask voting mechanism is designed to achieve accurate reconstruction of complex tree componentization structures through staged recognition and priority label allocation. This method performs well in multiple task computation evaluations, effectively avoids the problems of detail loss and semantic confusion in traditional modeling, and provides technical support for efficient 3D tree modeling and ecological applications.

In AR assembly, only experienced workers can complete the physical tasks with high operation precision requirements. The reason is that they use their brains to transform the information reflecting micro-operation intention into a series of manual operation rules. Because novices are not familiar with these rules, they often encounter problems of low precision and low efficiency. In order to describe these rules with AR instruction, this paper does some research work. Firstly, traditional AR and microAR are defined, and their differences and relations in guiding micro-operation are explained. On this basis, a microAR oriented to human microoperation rules is proposed. Then, according to the traditional AR and microAR, three instructions are designed to guide novice micro-operation assembly. An example is designed to test the assembly time and operation experience of three interfaces under a physical task. The results show that compared with traditional AR instruction, microAR instruction can better guide novice micro-operation.

The research of augmented reality physical tasks for human-computer collaboration is still an attractive dynamic research field. In recent years, AR Instruction supporting collaborative assembly has been developed, which can transfer graphic annotation between man and machine. However, there is little research on high precision manual assembly. This paper proposes a high-precision cognitive AR representation (M-AR) for manual operation guidance, which aims to transform a high-precision assembly relationship into a series of visual information to meet the user’s cognition of high-precision task intention. Using this representation, we conducted a within-subject case study to compare it with original AR representation (original AR) and low-precision cognitive AR representation. The results showed that there were significant differences in execution time and operation errors. The user’s subjective feedback also showed that M-AR supporting cognitive feedback can significantly improve a user’s cognitive experience in a physical task.

Research on Augmented Reality (AR)/Mixed Reality (MR) remote collaboration for physical tasks remains a compelling and dynamic area of study. AR systems have been developed which transmit virtual annotations between remote collaborators, but there has been little research on how haptic feedback can also be shared. In this paper, we present a Virtual Reality (VR)-Spatial Augmented Reality (SAR) remote collaborative system that provides haptic feedback with tangible interaction between a local worker and a remote expert helper. Using this system, we conducted a within-subject user study to compare two interfaces for remote collaboration between a local worker and expert helper, one with mid-air free drawing (MFD) and one with tangible physical drawing (TPD). The results showed that there were no significant differences with respect to performance time and operation errors. However, users felt that the TPD interface supporting passive haptic feedback could significantly improve the remote experts’ user experience in VR. Our research provides useful information on the way for gesture- and gaze-based multimodal interaction supporting haptic feedback in AR/MR remote collaboration on physical tasks.

Abstract This paper investigates the effect of using augmented reality (AR) annotations and two different gaze visualizations, head pointer (HP) and eye gaze (EG), in an AR system for remote collaboration on physical tasks. First, we developed a spatial AR remote collaboration platform that supports sharing the remote expert’s HP or EG cues. Then the prototype system was evaluated with a user study comparing three conditions for sharing non-verbal cues: (1) a cursor pointer (CP), (2) HP and (3) EG with respect to task performance, workload assessment and user experience. We found that there was a clear difference between these three conditions in the performance time but no significant difference between the HP and EG conditions. When considering the perceived collaboration quality, the HP/EG interface was statistically significantly higher than the CP interface, but there was no significant difference for workload assessment between these three conditions. We used low-cost head tracking for the HP cue and found that this served as an effective referential pointer. This implies that in some circumstances, HP could be a good proxy for EG in remote collaboration. Head pointing is more accessible and cheaper to use than more expensive eye-tracking hardware and paves the way for multi-modal interaction based on HP and gesture in AR remote collaboration.

This paper investigates how two different unique gaze visualizations (the head pointer(HP), eye gaze(EG)) affect table-size physical tasks in Mixed Reality (MR) remote collaboration. We developed a remote collaborative MR Platform which supports sharing of the remote expert's HP and EG. The prototype was evaluated with a user study comparing two conditions: sharing HP and EG with respect to their effectiveness in the performance and quality of cooperation. There was a statistically significant difference between two conditions on the performance time, and HP is a good proxy for EG in remote collaboration.

Controlling absolute magnitudes of fingertip force is an important skill in many haptic interactions such as surgical operations and mechanical assemblies. A fundamental question in the force control is how quickly human can output a target force with expected accuracy. In this paper, human's capability to control absolute magnitudes of fingertip force under audio or visual feedback was observed through experiments. Twelve participants applied a target force by pressing a force sensor with their fingers and maintained the force within a specified tolerance for a specific period of time (500ms). The magnitudes of target force and tolerance were manipulated systematically to yield varied levels of task difficulty. The experimental data showed that the applied fingertip force obeyed Fitts' law under both visual and auditory feedback modes when the index of difficulty (ID) was smaller than a threshold. The results may be used as guidelines for the applications that rely on accurate and quick changing force control over a target region such as multiple tapping tasks.

Due to lack of effective model representation, current MBD models cannot perform well in manufacturing reuse. In this paper, a new multilevel structuralized MBD model for manufacturing reuse is presented to capture the abstract information, topological information and machining semantic information involved. Firstly, the machining features considered as the machining semantic carriers are recognized from the MBD model. Then, the coupled machining feature cluster is introduced to construct the greater granularity structural elements than the machining features. Finally, the MBD model is structuralized by reorganizing the machining features and the coupled machining feature clusters in a hierarchical way. In the experiment, an aircraft structural part is utilized to verify our approach for manufacturing reuse.

This paper describes a novel method for retrieving 3D models. Following the principle of compound-eye vision, the proposed method represents a 3D model as a spherical image, and discriminates different 3D models using their corresponding spherical images. Meanwhile, by borrowing the concept of the Scale-Invariant Feature Transform (SIFT) algorithm, we design a feature extraction algorithm, named Spherical-SIFT, for extracting the salient local features on spherical images. Moreover, the Bag-of-Features approach is employed so as to achieve efficient comparison of different 3D models. The experimental results show the superior performance of our method over pervious methods.

The representation of triangular mesh surface is proposed using spherical image. Surface signals like normals and curvatures are stored in spherical images using spherical parameterization and partition of the sphere. The generation depends on several geometric signals, not one single geometric signal. So its color reflects the properties of the triangular mesh surface more clearly. With simple arrays similar to 2D image, the spherical image can be post-processed easily using sophisticated image processing algorithms. Applications include segmentation, editing and model retrieval.

This paper proposed to get 3D process model and final part through by process language understanding and graphics understanding. Based on the characteristics and structure of the machining process language, machining process language could be represented by ldquofour elementsrdquo. Then constructed the knowledge base based on ontology. Through dividing the process statements into five ldquoblocksrdquo and extracting the process semantic information, proposed a process language understanding method. At last, combined with the graphics understanding, we developed a prototype system and gave a case. The result showed that the method was effective.

Rapid growth and spread of information technology especially network technology enables people to use computers more naturally and easily and as a result, more design data and information leave in designers' desktops. Recent advances in grid, semantic Web, P2P and Web services have revolutionized the way we communicate and collaborate and also make it possible to integrate desktop data. Undertaking the intersection fulfillment of these technologies on enterprise desktop enables new integrative and collaborative use of not only organizational data but also distributed, autonomous personal data on the Web by fully leveraging potentially useful information sources on many desktops. In this context, we introduce SDDG (semantic desktop data grid) a P2P-semantic-grid enabled information sharing infrastructure for entire enterprise information, which wraps both organizational and personal information sources as semantic grid services and creates a semantic information interoperability environment following a P2P data coordination way. Through a discussion of fundamental scientific positions in general and approaches to information integration research in particular, a three-tier hybrid framework is presented and discussed in detail. The architecture for SDDG is an extension of our previous work ESD by reflecting principles of P2P data integration in semantic grid

Nowadays, swirling around most businesses is not structure information but semi-structure information, such as spreadsheets, and documents created by an individual or company. To fully manage and use semi-structure information as structure information does, we propose a technology based on the bi-directional integration between XML and RDB in a three-tiered structure. Through undertaking research at the intersection of XML and RDB, the technology expects to represent and use semi-structure information based on XML, and manage semi-structure information using RDB. Bi-directional means that we transform semi-structure data represented by XML into RDB and oppositely publishing RDB data into corresponding XML format based on schema mapping between XML and RDB, which is the first tier named static transformation tier in the three-tired structure. On the basis of static transformation tier, we transform semi-structure data query, i.e. query on XML from user interface into SQL on the corresponding bottom RDB and return back the result data in XML format, which is the second tier named dynamic query tier. At last, synchronous updating for the virtual XML and bottom RDB is carried out as the third tier named synchronous updating tier. All the three tiers' integration is realized through the middleware, which is also offered in this paper

Although a lot of previous work on schema matching has developed many partial automatic matches for specific application domains, combining multiple match techniques enables achieving high accuracy for a large variety of match circumstances. In this context, we present a schema-based element matching approach that concatenates linguistic-based matchers and a constraint-based matcher. We propose a basic processing of our element level match approach in terms of a sequence of linguistic-based match and constraint-based match. We also provide a composite element name matcher to automatically combine linguistic-based match algorithms with a maximum priority strategy, and a neural network matcher to categorize elements of schemas by using element constraints with results from composite name matcher for joint consideration of multiple criteria. The concatenation of composite name matcher and neural network matcher enable our approach to adapt to more complex matching circumstance

A novel method for recognizing multifont Chinese characters is presented. After character skeleton extraction, a classification is first made according to a new complexity measure: a weighted sum of endpoints and nodes. Unlabeled feature point matching between an input character and model characters is realized by minimizing an approximate Euclidean distance. The final recognition decision is based on calculating a flexible similarity function which depends on the feature point dispersion of the considered fonts. Experiments on a database of 1000 Chinese characters have been conducted. The recognition rate exceeds 96%. >