scispace - formally typeset
Search or ask a question

Showing papers on "3D single-object recognition published in 2019"


Journal ArticleDOI
TL;DR: It is possible to perform scale-invariant 3D object recognition and achieve a high recognition rate when combined with the feature point selection algorithm proposed in this study by using the gradients of the scalar functions defined on the 3D surface.
Abstract: As 3D scanning technology develops, it becomes easier to acquire various 3D surface data; thus, there is a growing need for 3D data registration and recognition technology. Many existing studies use local descriptors using local surface patches, and most of them use a fixed support radius, so they cannot cope perfectly when the model and scene have different scales. In this study, we propose a perfectly scale-invariant feature selection algorithm by extending the 2D SIFT algorithm (Lowe in Int J Comput Vis 60(2):91–110, 2004) to a 3D mesh. The feature selection method proposed in this study can obtain highly repeatable feature points and support radii regardless of mesh scale. The selected features can effectively describe the local information by the new shape descriptor proposed in this study. Unlike existing shape descriptors, it is possible to perform scale-invariant 3D object recognition and achieve a high recognition rate when combined with the feature point selection algorithm proposed in this study by using the gradients of the scalar functions defined on the 3D surface. We also reduced the searching space and lowered the false positive rate by suggesting a new RANSAC-based transformation hypotheses generation algorithm. Our 3D object recognition algorithm achieves recognition rates of 100 and 98.5%, respectively, when tested on the U3OR and CFVD datasets, exceeding the results of previous studies.

9 citations


Journal ArticleDOI
TL;DR: A genetic algorithm-based method based on geometric features that can solve the combinatorial optimization problem effectively and can accurately distinguish the contour of the target object from the background is developed.
Abstract: Object recognition in complex backgrounds has challenged the fields of pattern recognition for years. It is even harder when the targets in images are of different poses. Current methods use descriptors of characteristic vectors and machine learning algorithms to produce classifiers for object recognition. However, the generalization ability of these methods relies on the quality of the training phase and cannot find the precise boundaries of the targets. The geometric features of objects are the most stable and consistent features, so the recognition method based on shapes can be more intuitive than those based on color and texture features. This paper proposes a novel method that uses the images represented by line segments. The recognition mission becomes to effectively filter and combine them. The contour fragments after combinations are expected to satisfy the given model, or certain parts of it. In this way, object recognition can be viewed as a combinatorial optimization problem. This paper develops a genetic algorithm-based method to solve this problem. The experimental results show that this method can solve the combinatorial optimization problem effectively and can accurately distinguish the contour of the target object from the background. This method, which is based on geometric features, may contribute to the development of explicit principals for the description of object structure and recognition method based on symbolic reasoning.

4 citations