3D object recognition and classification: a systematic literature review
Citations
438 citations
101 citations
17 citations
16 citations
9 citations
References
46,906 citations
"3D object recognition and classific..." refers background or methods in this paper
...The local features are extracted via SIFT chain features, which are employed for subspace construction through PCA....
[...]
..., in the feature-based representation there is a variety of feature descriptors (SHOT [288], SI[116], VFH [254], SIFT [175] and so on)....
[...]
...The proposed method is composed of two stages: the offline stage, where the Bundler motion method structure is applied into the object data set and the background point are manually removed, to obtain the object point cloud model and generate, from this model, the aspect graph aware representation, and the online stage, where coarse 2D–3D correspondences are produced, by similarity computation between SIFT descriptors from input image and 3D model, and refined via a two-stage filter application, which removes false correspondences....
[...]
...In the recognition process, candidate evidences include global and local features extracted from the RGB-D data, i.e., the 3D SIFT, CLB and shape descriptors extracted from the point cloud....
[...]
...The traditional pooling in the space domain cannot be applied directly to high-dimensional pooling domains such as SIFT and FPFH due to the exponential pooling bins growth number....
[...]
42,067 citations
16,989 citations
"3D object recognition and classific..." refers methods in this paper
...The scale-invariant feature transform (SIFT), presented by Lowe [174], used for describing salient points (keypoints) and representing the objects, was employed in several analyzed works as a form to extract keypoints....
[...]
13,993 citations
"3D object recognition and classific..." refers methods in this paper
...Using this theory and the steps presented in [112] and [98] the 3D keypoint detector, BIK-BUS is generated....
[...]
10,525 citations