Detection and Handling of Dynamic Scenes During an Active Vision Process for Object Recognition Using a Boundary Representation

This paper present a new method for automatic form feature recognition from boundary representation of a solid model. The proposed method is called the edge boundary classification technique. This technique uses the geometry, topology, and spatial information of a solid model for feature recognition patterns. Some aspects of edge boundary classification are discussed for recognising non-intersecting and intersecting form features.

Recently, concurrent engineering (CE) has become one of the major conceptual approaches in reducing product developing time. Design for manufacturing (DFM) and design for assembly (DFA) are two of the major areas in CE. Several researchers have integrated a CAD/CAM system and a knowledge-based system to implement DFA and DFM concepts. However, the incompatibility of the data representation methods between CAD/CAM and knowledge-based systems is the major issue in this integration method. In this paper, an approach is proposed to find the minimum wall thickness of a part from its boundary representation (B-Rep) model. These data can then be directly utilized by a knowledge-based system to analyse the manufacturability of thin wall parts such as injection moulded or diecast parts.

Loops are critical elements in boundary representation (B-rep) models because they link all edges corresponding to a face. Loops can be used in feature recognition for identifying depressions or protrusions. In real 3D CAD models, however, features typically lie across multiple faces, which is beyond the data structure of current B-rep models. This study presents a virtual loop concept to account for all loop types used in CAD models, and develops algorithms for recognizing them. In accordance with the complexity of the recognition algorithm, this study defines three types of loop: single, virtual, and multivirtual. A single loop is the current loop recorded in the B-rep model. A virtual loop lies across faces that are at least G1 continuous. Finally, a multivirtual loop lies across faces that are either G0 or G1 continuous. The proposed loop structure provides a more complete data structure for recognizing various types of features in feature-recognition modules. Several realistic CAD models are presented to confirm the feasibility of the proposed loop-recognition and feature-recognition methods.

An automatic generation approach of process model based on feature knowledge and geometric modeling

Defects in boundary representation models often lead to system errors in modeling software and associated applications. This paper analyzes the model rectification problem of manifold boundary models, and argues that a rectify-by-reconstruction approach is needed in order to reach the global optimal solution. The restricted face boundary reconstruction problem is shown to be NP-hard. Based on this, the solid boundary reconstruction problem is also shown to be NP-hard.

After analyzing the characteristics of rock salt and the feasibility and importance of using underground rock salt cavern as energy stockpiles, this paper explains in detail about the principle of sonar detection technology in cavern measuring and its data organization. This paper studies the technology of three-dimensional surface topological reconstruction of underground rock salt cavern based on sonar detection data with half-edge data structure and boundary representation models. Meanwhile, by conducting validity check for constructed surface model and repair of possible defects, the paper finds correct and effective three-dimensional surface models of underground rock salt cavern, and provides effective data base for the stability analysis, creep analysis and other numerical simulations of underground rock salt cavern. This method, with important economic and social significance, can provide a scientific basis and technological support for the construction of caverns, especially for the safe and rational use of underground rock salt cavern.

NC machine tool path generation from CSG part representations

The design of a mould, which is manually performed, is time-consuming and quite complicated. This research presents a study on the procedures for automatic feature recognition to help in mould design. A hybrid representation approach is used in automatic feature recognition to extract geometric information from a feature to identify undercut features. Boundary representation is applied in the shape representation using topology and geometry. The proposed approach uses a face adjacency hypergraph to describe the shape of the undercut features by representing the relationships amongst the feature faces. Face-to-face composition is applied to represent the relationship between the main and the undercut features. Parts A, B, and C are created, and automatic feature recognition is used to classify the depression and protrusion features. Algorithms based on a heuristic rule are adopted to determine the optimal parting line and direction of the parts by comparing their feature’s geometric information. The automatic feature recognition approach identifies the shape of concave and convex features in accordance with the depression or protrusion of faces and the face adjacency relationship. This approach can be helpful in automatically creating the core and cavity. The proposed approach will help simplify the process and reduce the time for mould design. Therefore, this approach will significantly affect the increase in productivity in the manufacturing industry.

Abstract. The applications and understanding of Land-use Regulations (LuR) are more communicable when they are linked to the digital representation of the physical world. In order to support issuing a planning permit and move towards the establishment of automated planning permit checks, this paper investigates how LuRs related to a planning permit process can be modelled in 3D called 3D CityLuR. 3D CityLuR serves as a 3D model for representing LuRs’ legal extents on a city scale. It is formed based on multiple geometric modelling approaches representing LuRs, which can provide a better cognitive understanding of LuRs and subsequently facilitate LuR automatic checks. To this purpose, according to LuRs’ descriptions and characteristics explained in related planning documents, key parameters representing LuRs’ extent are identified (e.g. maximum distance in overlooking or maximum allowed height in building height regulations). Accordingly, to automatically model each LuR, a geometric modelling approach (e.g. Boundary Representation (B-Rep), CSG, and extrusion) that best fits with the identified key parameters is proposed. In addition, to combine 3D CityLuR with an integrated BIM-GIS environment, the level of information need in terms of geometries and semantics is specified. Finally, the paper results in a showcase for five LuRs including building height, energy efficiency protection, overshadowing open space, overlooking, and noise impacts regulations. The showcase is a proof of concept for determining how these LuRs can be modelled in 3D and combined with 3D city models based on the selected geometric modelling approaches, identified parameters, and level of information need.

Fetuses are extremely radiosensitive and the protection of pregnant females against ionizing radiation is of particular interest in many health and medical physics applications. Existing models of pregnant females relied on simplified anatomical shapes or partial-body images of low resolutions. This paper reviews two general types of solid geometry modeling: constructive solid geometry (CSG) and boundary representation (BREP). It presents in detail a project to adopt the BREP modeling approach to systematically design whole-body radiation dosimetry models: a pregnant female and her fetus at the ends of three gestational periods of 3, 6 and 9 months. Based on previously published CT images of a 7-month pregnant female, the VIP-Man model and mesh organ models, this new set of pregnant female models was constructed using 3D surface modeling technologies instead of voxels. The organ masses were adjusted to agree with the reference data provided by the International Commission on Radiological Protection (ICRP) and previously published papers within 0.5%. The models were then voxelized for the purpose of performing dose calculations in identically implemented EGS4 and MCNPX Monte Carlo codes. The agreements of the fetal doses obtained from these two codes for this set of models were found to be within 2% for the majority of the external photon irradiation geometries of AP, PA, LAT, ROT and ISO at various energies. It is concluded that the so-called RPI-P3, RPI-P6 and RPI-P9 models have been reliably defined for Monte Carlo calculations. The paper also discusses the needs for future research and the possibility for the BREP method to become a major tool in the anatomical modeling for radiation dosimetry.For more information on this article, see medicalphysicsweb.org

Formal specification of topological subdivisions using hypermaps

Implicit Surface-Based Geometric Fusion

An efficient approach is proposed to the design of a kinematically feasible path in the presence of obstacles, based on task requirements. First, a path is planned by specifying both position and orientation for each point along the path. Then, suitable expressions are derived to describe the relationship between the planned path and the robot workspace. With the geometric information from these expressions and the modification strategies based on different task requirements, the infeasible portions due to the kinematic constraints and obstacles are detected and modified accordingly. A six revolute joint robot manipulator and a sphere obstacle are selected as the case study. The results are further extended to general types of industrial robot manipulators. >

We propose a novel approach for multi-view object detection in 3D scenes reconstructed from RGB-D sensor. We utilize shape based representation using local shape context descriptors along with the voting strategy which is supported by unsupervised object proposals generated from 3D point cloud data. Our algorithm starts with a single-view object detection where object proposals generated in 3D space are combined with object specific hypotheses generated by the voting strategy. To tackle the multi-view setting, the data association between multiple views enabled view registration and 3D object proposals. The evidence from multiple views is combined in simple bayesian setting. The approach is evaluated on the Washington RGB-D scenes datasets [1], [2] containing several classes of objects in a table top setting. We evaluated our approach against the other state-of-the-art methods and demonstrated superior performance on the same dataset.

RGB-D object detection has achieved significant advance, because depth provides complementary geometric information to RGB images. Considering depth images are unavailable in some scenarios, we focus on depth privileged object detection in indoor scenes, where the depth images are only available in the training phase. Under this setting, one prevalent research line is modality hallucination, in which depth image and depth feature are the common choices for hallucinating. In contrast, we choose to hallucinate depth deformation, which is explicit geometric information and efficient to hallucinate. Specifically, we employ the deformable convolution layer with augmented offsets as our deformation module and regard the offsets as geometric deformation, because the offsets enable flexibly sampling over the object and transforming to a canonical shape for ease of detection. In addition, we design a quality-based mechanism to avoid negative transfer of depth deformation. Experimental results and analyses on NYUDv2 and SUN RGB-D demonstrate the effectiveness of our method against the state-of-the-art methods for depth privileged object detection.