Paper Info
Title
Metrological Evaluation Of Modeling Method Between Kinect Fusion And 3d Scan With Kinect V2
Abstract
In this paper an experimental study of real-time modeling method with Kinect Fusion and 3D Scan respectively was carried out from the visual effect and the accuracy of scanning to obtain point cloud data. Cauliflower, pitaya and carrot, which size, shape, and surface roughness are all different, were selected as the research crops and Kinect V2 (Kinect for windows V2) developed by Microsoft was selected as the sensor. The acquisition environment was designed and processed to remove a lot of noise points around the scene. The experimental results show that the visual effect is the best and the modeling time is the shortest with Kinect Fusion method when the distance between object and Kinect is 83.5cm, VMIW is 200, and VVPM is 512. According to the glob al fluctuations at different angles of the three crop s' point cloud data, it can be concluded that the accuracy of point cloud data is higher, and the details of the objects is more complete and detailed obtained with Kinect Fusion method than with 3D Scan method. However, with the improvement of surface roughness the capability of modeling method is reduce. The proposed modeling method based on Kinect V2 can be used as reference for other crops.
Year
DOI
Venue
2018
10.1145/3195106.3195149
PROCEEDINGS OF 2018 10TH INTERNATIONAL CONFERENCE ON MACHINE LEARNING AND COMPUTING (ICMLC 2018)
Keywords
Field
DocType
Machine vision, Crops, Point cloud data, Modeling method
Computer vision,Machine vision,Pattern recognition,Computer science,Metrology,Fusion,Artificial intelligence,Point cloud,Surface roughness
Conference
Citations 
PageRank 
References 
0
0.34
0
Authors
3
Name
Order
Citations
PageRank
Xia-Xia Guo100.34
Gui-Hong Zhou200.34
Hong Cheng370365.27