• ETRI Journal
• /
• v.42 no.1
• /
• pp.78-89
• /
• 2020

Human activity recognition (HAR) has become effective as a computer vision tool for video surveillance systems. In this paper, a novel biometric system that can detect human activities in 3D space is proposed. In order to implement HAR, joint angles obtained using an RGB-depth sensor are used as features. Because HAR is operated in the time domain, angle information is stored using the sliding kernel method. Haar-wavelet transform (HWT) is applied to preserve the information of the features before reducing the data dimension. Dimension reduction using an averaging algorithm is also applied to decrease the computational cost, which provides faster performance while maintaining high accuracy. Before the classification, a proposed thresholding method with inverse HWT is conducted to extract the final feature set. Finally, the K-nearest neighbor (k-NN) algorithm is used to recognize the activity with respect to the given data. The method compares favorably with the results using other machine learning algorithms.

• Journal of Industrial Technology
• /
• v.38 no.1
• /
• pp.41-52
• /
• 2018

Recently, object detection is a critical function for any system that uses computer vision and is widely used in various fields such as video surveillance and self-driving cars. However, the conventional methods can not detect the objects clearly because of the dynamic background change in the beach. In this paper, we propose a new technique to detect humans correctly in the dynamic videos like shores. A new background modeling method that combines spatial GMM (Gaussian Mixture Model) and temporal GMM is proposed to make more correct background image. Also, the proposed method improve the accuracy of people detection by using SVM (Support Vector Machine) to classify people from the objects and KCF (Kernelized Correlation Filter) Tracker to track people continuously in the complicated environment. The experimental result shows that our method can work well for detection and tracking of objects in videos containing dynamic factors and situations.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.65 no.3
• /
• pp.477-486
• /
• 2016

This paper presents the area classification, identification, and tracking for multiple moving objects with the similar colors. To do this, first, we use the GMM(Gaussian Mixture Model)-based background modeling method to detect the moving objects. Second, we propose the use of the binary and morphology of image in order to eliminate the shadow and noise in case of detection of the moving object. Third, we recognize ROI(region of interest) of the moving object through labeling method. And, we propose the area classification method to remove the background from the detected moving objects and the novel method for identifying the classified moving area. Also, we propose the method for tracking the identified moving object using Kalman filter. To the end, we propose the effective tracking method when detecting the multiple objects with the similar colors. Finally, we demonstrate the feasibility and applicability of the proposed algorithms through some experiments.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.8 no.11
• /
• pp.1647-1654
• /
• 2013

Detection performance of the vehicle on the road depends on weather conditions, the shadow by the movement of the sun, or illumination changes, etc. In this paper, a vehicle detection system in conjunction with a robust background estimate algorithm to environment change on the road in daytime is proposed. Gaussian Mixture Model is applied as background estimation algorithm, and also, Adaboost algorithm is applied to detect the vehicle for candidate region. Through the experiments with input videos obtained from a various weather conditions at the same actual road, the proposed algorithm were useful to detect vehicles in the road.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2011.10a
• /
• pp.157-160
• /
• 2011

In this paper, We proposed the method of recognition and analysis of human activites using Motion vector in real-time surveillance system. We employs subtraction image techniques to detect blob(human) in the foreground. When MPEG-4 video recording EPZS(Enhanced Predicted Zonal Search) is detected the values of motion vectors were used. In this paper, the activities of human recognize and classified such as meta-classes like this {Active, Inactive}, {Moving, Non-moving}, {Walking, Running}. Each step was separated using a step-by-step threshold values. We created approximately 150 conditions for the simulation. As a result, We showed a high success rate about 86~98% to distinguish each steps in simulation image.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2003.05a
• /
• pp.47-50
• /
• 2003

최근 IT분야에서 주목받고 있는 부분은 정보가전분야이다. 특히, 홈오토메이션(HA Home Automation)에서 중심이 되는 시스템은 홈서버라 할 수 있다. 즉, 가정내에 홈서버가 설치되어 이를 중심으로 내부적으로는 정보가전기기들이 네트워킹되어 있고, 밖으로는 인터넷을 기본 통신 환경으로 구성되는 것이 현실이다. 이러한 환경에서 홈시큐리티 방면의 영상감시기능은 핵심분야 중 하나이다. 영상감시기능을 수행하고자 가정의 홈서버상에 영상감시 모듈이 탑재되어진다. 이때, 홈서버는 임베디드 시스템으로 Qplus라고 하는 임베디드 리눅스 계열의 운영체제를 기본으로 하며, 영상감시 모듈은 SIP(Session Initiation Protocol)기반으로 수행된다. SIP은 VoIP(Voice over IP) 분야의 핵심기술로 최근 많이 부각되어 널리 활용되고 있는 응용 계층의 시그널링 프로토콜이다. 한편, 영상코덱은 ITU-T의 H.261표준을 따르고 있으며, 이러한 영상감시기능은 홈서버 뿐만 아니라, PDA와 같은 핸드헬드 장치를 통해서도 제공된다. 본 논문에서는 임베디드 영상가미 시스템의 설계 및 구현에 대해 기술하고 있다.

• Journal of Satellite, Information and Communications
• /
• v.12 no.2
• /
• pp.56-60
• /
• 2017

This paper describes the design and implementation of fall-down detection algorithm based on image processing. The fall-down detection algorithm separates objects by using background subtraction and binarization after grayscale conversion of the input image acquired by the camera, and recognizes the human body by using labeling operation. The recognized human body can be monitored on the display image, and an alarm is generated when fall-down is detected. By using computer simulation, the proposed algorithm has shown a detection rate of 90%. We verify the feasibility of the proposed system by verifying the function by using the prototype test implemented on the DSP image processing board.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.18 no.3
• /
• pp.422-428
• /
• 2008

In this paper, we propose the robust object detection algorithm with spatial gradient information. To do this, first, we eliminate error values that appear due to complex environment and various illumination change by using prior methods based on hue and intensity from the input video and background. Visible shadows are eliminated from the foreground by using an RGB color model and a qualified RGB color model. And unnecessary values are eliminated by using the HSI color model. The background is removed completely from the foreground leaving a silhouette to be restored using spatial gradient and HSI color model. Finally, we validate the applicability of the proposed method using various indoor and outdoor conditions in a complex environments.

• Journal of Korea Multimedia Society
• /
• v.22 no.2
• /
• pp.136-145
• /
• 2019

The crowded environment of a domestic pig farm is highly vulnerable to the spread of infectious diseases such as foot-and-mouth disease, and studies have been conducted to automatically analyze behavior of pigs in a crowded pig farm through a video surveillance system using a camera. Although it is required to correctly separate occluding pigs for tracking each individual pigs, extracting the boundaries of the occluding pigs fast and accurately is a challenging issue due to the complicated occlusion patterns such as X shape and T shape. In this study, we propose a fast and accurate method to separate occluding pigs not only by exploiting the characteristics (i.e., one of the fast deep learning-based object detectors) of You Only Look Once, YOLO, but also by overcoming the limitation (i.e., the bounding box-based object detector) of YOLO with the test-time data augmentation of rotation. Experimental results with two-pigs occlusion patterns show that the proposed method can provide better accuracy and processing speed than one of the state-of-the-art widely used deep learning-based segmentation techniques such as Mask R-CNN (i.e., the performance improvement over Mask R-CNN was about 11 times, in terms of the accuracy/processing speed performance metrics).

• Fire Science and Engineering
• /
• v.27 no.5
• /
• pp.15-18
• /
• 2013

This paper is concerned with the Performance and Reliability Certification for fire detection system in outdoor area such small and middle sized cultural assets, natural monument and outdoor facilities. Especially, if a fire were to occur in vulnerable area, it is difficulty to detect a fire. therefore we propose a high efficiency and low cost unmanned fire detection system in capable of an early detection regardless spontaneously fire or firebug. for Adoption of Intelligent Fire Detection System with movable and unmanned function breaking from the existing Conventional Fire Detection System, this Range of R&D includes the Performance test, Function test, Field test, Flame Detection test and EMI/EMS Compliance test. the Result data of Performance test, Function test and Field test is generally good during 3 months. also we checked that thermal variation test and EMI/EMS compliance test are good result data within allowable range. As a result of general test, we verified improvement results that the measure distance of fire detection extend 75 m, the Power of waiting time increase 4 hours, the Power of operation time increase 3 days and the context awareness with video as well as sensors.