|
Band Selection for Plastic Classification using NIR Hyperspectral Image |
|
Heekang Kim, Sungho Kim(Yeungnam University, Korea) |
| Recently, there are many plastics classification methods, because the importance of recycling increases. In case of plastic, optical methods such as Near infrared (NIR) hyperspectral image sensors were introduced. But, hyperspectral image have hundreds of bands, therefore the calculation time increase. This paper presents a band selection method to reduce the calculation time for plastic classification to solve this problem. |
|
|
Multiple Threading Strategy for Intelligent Surveillance System |
|
Wahyono Wahyono, Kang-Hyun Jo(University of Ulsan, Korea) |
| In recent years, an intelligent surveillance system is widely developed along with the rapid use of closed circuit television (CCTV) camera in the public area. The system should be able to process multiple source data from different CCTV networks simultaneously and automatically detect any suspicious event on the monitoring area. A low-cost computational time is required to make a fast-enough system in order to prevent any delay in decision-making of the security officer. In this paper, multiple threading based process is proposed to achieve this requirement. |
|
|
Efficient Object Recognition for Industrial Inspection |
|
Jungho Kim(Korea Electronics Technology Institute, Korea), Youngbae Hwang(KETI, Korea), Wang-Heon Lee(Hansei University, Korea) |
| In this paper, we present an efficient method of recognizing and inspecting objects for product inspection. For this purpose, we employ one of binary features, ORB which shows both good recognition rate and fast extraction time. For fast matching, we introduce how to construct the vocabulary tree for binary features using k-mediods clustering. In addition, we present a method of deciding defective products using Markov Random Fields(MRF)-based optimization. Using the KAIST-104 database, we demonstrate fast recognition time and feasibility for product inspection. |
|
|
Extrinsic calibration of a camera and laser range finder |
|
Nam-Hun Kim, Jong-Eun Ha(Seoul National University of Science and Technology, Korea) |
| Extrinsic calibration between a camera and laser range finder is necessary to fuse information from both
sensor. In this paper, we propose an algorithm for the extrinsic calibration of a camera and laser range finder by using a
new structure that is consisted of multiple vertical planes on the ground plane. Experimental results show the feasibility
of proposed algorithm. We compare projected points to actual laser points on image. |
|