No abstract

No abstract

While benchmark datasets have been proposed for testing computer vision and 3D shape retrieval algorithms, no such datasets have yet been put forward to assess the relevance of these techniques for engineering problems. This paper presents several distinctive benchmark datasets for evaluating techniques for automated classification and retrieval of CAD objects. These datasets include (1) a dataset of CAD primitives (such as those common in constructive solid geometry modeling); (2) two datasets consisting of classes generated by minor topological variation; (3) two datasets of industrial CAD models classified based on object function and manufacturing process, respectively; (4) and a dataset of LEGO c models from the Mindstorms c robotics kits. Each model in the datasets is available in three formats -ACIS SAT, ISO STEP, and as a VRML mesh (some models are available under several different fidelity settings). These are all available through the National Design Repository.Using these datasets, we present comprehensive empirical results for nine (9) different shape and solid model matching and retrieval techniques. These experiments show, as expected, that the quality of precision-recall performance can significantly vary on different datasets. These experiments reveal that for certain object classes and classifications, such as those based on manufacturing processes, all existing techniques perform poorly. This study reveals the strengths and weaknesses of existing research in these areas, introduces open challenge problems, and provides meaningful datasets and metrics against which the success of current and future work can be measured.

This paper describes a new approach to automate the classification of solid models using machine learning techniques. Existing approaches, based on group technology, fixed matching algorithms or pre-defined feature sets, impose a priori categorization schemes on engineering data or require significant human labeling of design data. This paper describes a shape learning algorithm and a general technique for "teaching" the algorithm to identify new or hidden classifications that are relevant in many engineering applications. In this way, the core shape learning algorithm can be used to find a wide variety of model classifications based on user input and training data. This allows for great flexibility in search and data mining of engineering data.

The ability to search for a CAD model that represents a specific physical part is a useful capability that can be used in many different applications. This paper presents an approach to use partial 3D point cloud of an artifact for retrieving the CAD model of the artifact. We assume that the information about the physical parts will be captured by a single 3D scan that produces dense point clouds. CAD models in our approach are represented as polygonal meshes. Our approach involves segmenting the point cloud and CAD mesh models into surface patches. The next step is to identify corresponding surface patches in point clouds and CAD models that could potentially match. Finally, we compute transformations to align the point cloud to the CAD model and compute distance between them. We also present experimental results to show that our approach can be used to retrieve CAD models of mechanical parts. INTRODUCTIONThe ability to search for a CAD model that represents a specific physical part is a useful capability that can be used in many different applications. The following scenario illustrates the usefulness of being able to search for a CAD model based on point cloud generated by a partial scan. Let us assume that a part needs to be replaced in a complex machine. There is no label on the part. Hence the user does not know the part number. The user scans the physical part using a 3D scanner and generates the point cloud. This point cloud is then used by the user to search the CAD database and find the CAD model of this part. The CAD model has the information about the part number and the user is able to order the replacement part using the part number.

Fig. 1. We explore a volume dataset with an intensity-gradient histogram segmentation hierarchy. 1. We compute a 2D intensitygradient histogram from a volume dataset. 2. We mimic the user visual search of shapes in the histogram by recursively segmenting the histogram image using normalized cuts. 3. We construct a multi-resolution hierarchy for interactive exploration. 4. Users traverse this hierarchy to discover features in the volume data and compose meaningful visualizations.Abstract-Visual exploration of volumetric datasets to discover the embedded features and spatial structures is a challenging and tedious task. In this paper we present a semi-automatic approach to this problem that works by visually segmenting the intensitygradient 2D histogram of a volumetric dataset into an exploration hierarchy. Our approach mimics user exploration behavior by analyzing the histogram with the normalized-cut multilevel segmentation technique. Unlike previous work in this area, our technique segments the histogram into a reasonable set of intuitive components that are mutually exclusive and collectively exhaustive. We use information-theoretic measures of the volumetric data segments to guide the exploration. This provides a data-driven coarse-to-fine hierarchy for a user to interactively navigate the volume in a meaningful manner.

This paper describes a new approach to automate the classification of solid models using machine learning techniques. Existing approaches, based on group technology, fixed matching algorithms or pre-defined feature sets, impose a priori categorization schemes on engineering data or require significant human labeling of design data. This paper describes a shape learning algorithm and a general technique for "teaching" the algorithm to identify new or hidden classifications that are relevant in many engineering applications. In this way, the core shape learning algorithm can be used to find a wide variety of model classifications based on user input and training data. This allows for great flexibility in search and data mining of engineering data.

No abstract