Abstract—In this paper, we introduce 2D view-based method
by using the combined depth image of the dice-shape flat
pattern, scale-invariant feature transform (SIFT) algorithm
and random sample consensus (RANSAC) algorithm. The
combined depth images are generated by the direction of a
square face views and these features are composed in the
feature DB. We previously did not use the combined depth
image of the dice-shape flat pattern but that of the tile-shape.
We compare the combined depth image of the dice-shape flat
pattern with that of the tile-shape. The combined depth images
of the dice-shape flat pattern are composed in the 6 depth
images and also that of the tile-shape. In the experiment, we use
the number of 16 classes in the SHREC benchmark database
and there are 3 to 4 the similar 3D models in the class.
Index Terms—Dice-shape flat pattern, SIFT, RANSAC,
tile-shape, depth image, 3D model identification.
Jeongseok Jo and Jongweon Kim are with the Copyright Protection
Department, Sangmyung University, 20, Hongjimun-2gil, Jongno-gu, Seoul,
03016, Korea (e-mail: jeongs329@naver.com, jwkim@smu.ac.kr).
Yongbae Kim is with the Creative Content Labs, Sangmyung University,
20, Hongjimun-2gil, Jongno-gu, Seoul, 03016, Korea (e-mail:
ybkimdb@gmail.com).
[PDF]
Cite: Jeongseok Jo, Yongbae Kim, and Jongweon Kim, "3D Model Identification Using the Dice-Shape Flat Pattern," International Journal of Knowledge Engineering vol. 2, no. 1, pp. 35-38, 2016.