Shape-based object retrieval by contour segment matching

Yang, Cong; Tiebe, Oliver; Pietsch, Pit; Feinen, Christian; Kelter, Udo; Grzegorzek, Marcin

doi:10.1109/icip.2014.7025446

Cited by 18 publications

(3 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the binary image classification, the authors of (Shekar and Pilar, 2014) use local morphological and binary patterns for the classification of objects. The authors of (Shen et al, 2014) use the contours associated with the skeleton of a binary image and the authors of (Yang et al, 2014) uses a contour segment matching to measure the similarity of shapes for object retrieval, which are segmented by a skeletonization method. Due to the rapid development in machine learning, object retrieval tasks for 2D images can be solved using for example, neuronal networks, support vector machines, or random forest (Bansal et al, 2020).…”

Section: Iced21mentioning

confidence: 99%

“…Such skeleton-based reverse engineering can be applied for deformed FE-meshes represented in (Louhichi et al, 2015), point cloud reconstruction (Kresslein et al, 2018), the determination of center lines in CT-Data (Computer Tomography) to reconstruct blood vessels (Hua Li and Yezzi, 2006) or vascular skeletons (Lidayová et al, 2016) or topology optimization results (Nana et al, 2017;Yin et al, 2020). Additional to the reverse engineering task, such classification can also be used for object retrieval tasks (Yang et al, 2014) to find for example, specific parts in 2D technical drawings of various amount of data sets. For reverse engineering, the parametrization of that cross-section, such as the radius or the size parameter of the cap and the type of cross-section particular for junctions, is required.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Euclidian Distance Transformation, Main Axis Rotation and Noisy Dilitation Supported Cross-Section Classification With Convolutional Neural Networks

et al. 2021

View full text Add to dashboard Cite

Polygon meshes and particularly triangulated meshes can be used to describe the shape of different types of geometry such as bicycles, bridges, or runways. In engineering, such polygon meshes can be supplied as finite element meshes, resulting from topology optimization or from laser scanning. Especially from topology optimization, frame-like polygon meshes with slender parts are typical and often have to be converted into a CAD (Computer-Aided Design) format, e.g., for further geometrical detailing or performing additional shape optimization. Especially for such frame-like geometries, CAD designs are constructed as beams with cross-sections and beam-lines, whereby the cross-section is extruded along the beam-lines or beam skeleton. One major task in the recognition of beams is the classification of the cross-section type such as I, U, or T, which is addressed in this article. Therefore, a dataset consisting of different cross-sections represented as binary images is created. Noisy dilatation, the distance transformation, and main axis rotation are applied to these images to increase the robustness and reduce the necessary amount of samples. The resulting images are applied to a convolutional neuronal network.

show abstract

Section: Iced21mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Euclidian Distance Transformation, Main Axis Rotation and Noisy Dilitation Supported Cross-Section Classification With Convolutional Neural Networks

et al. 2021

View full text Add to dashboard Cite

show abstract

“…In contrast, our proposal does not involve any learning algorithm. Contour segment matching (Yang et al, 2014) and moment invariants (Premaratne and Premaratne, 2014) have also been proposed for the matching step. In these methods, the query image needs to be compared with every image in the database.…”

Section: Related Workmentioning

confidence: 99%

Skewness map: estimating object orientation for high speed 3D object retrieval system

Sintunata

Kurumi

Aoki

2018

IJIEI

View full text Add to dashboard Cite

3D object retrieval system is a system where a similar or the same object in the database should be retrieved given a 2D query image (sketches or photographs). Unfortunately, as the appearance of 3D object might vary depending on the viewing directions, a vast amount of 2D rendered images must be processed (matched) to solve this problem. In this paper, we present a novel method called skewness map to relieve this problem. Skewness map can estimate the orientation of the object and select a few representative images accurately from the database; therefore matching every image in the database can be avoided. Experimental results show the retrieval system becomes much faster (14 times faster in matching time) and accurate in estimating the object orientation (less than one-degree error in average).

show abstract