Seok-Kap Ko

Design and implementation of a video surveillance system for a Sensor Network application service

Object movement detection is key component of remote surveillance and monitoring system. In this paper, we designs a reliable moving object detection system for video surveillance which works well in a Ubiquitous Sensor Networks (USN) environment. Our system has been implemented using Open Geospatial Consortium (OGC) Sensor Web Enablement (SWE) standard, and operated in the android platform of mobile environment. As a result, our system can report on object detection and object counting information to the management server in real-time.

Object classification of UWB responses using S T-CNN

UWB response includes unique characteristics of reflecting objects. Because the response is the combination of many distortion, resonance, and multi-paths, the object classification of UWB response is difficult. In this paper, we propose an object classification method using S-transform and convolution neural network. S-transform converts time series data of UWB response to frequency-time domain which convolutional neural network can learn and classify.