Sung-Kwan Kang

Development of head detection and tracking systems for visual surveillance

This paper proposes a technique for the detection of head nod and shake gestures based on eye tracking and head motion decision. The eye tracking step is divided into face detection and eye location. Here, we apply a motion segmentation algorithm that examines differences in moving people’s faces. This system utilizes a Hidden Markov Model-based head detection module that carries out complete detection in the input images, followed by the eye tracking module that refines the search based on a candidate list provided by the preprocessing module. The novelty of this paper is derived from differences in real-time input images, preprocessing to remove noises (morphological operators and so on), detecting edge lines and restoration, finding the face area, and cutting the head candidate. Moreover, we adopt a K-means algorithm for finding the head region. Real-time eye tracking extracts the location of eyes from the detected face region and is performed at close to a pair of eyes. After eye tracking, the coordinates of the detected eyes are transformed into a normalized vector of x-coordinate and y-coordinate. Head nod and shake detector uses three hidden Markov models (HMMs). HMM representation of the head detection can estimate the underlying HMM states from a sequence of face images. Head nod and shake can be detected by three HMMs that are adapted by a directional vector. The directional vector represents the direction of the head movement. The vector is HMMs for determining neutral as well as head nod and shake. These techniques are implemented on images, and notable success is notified.

Bio-Interactive Healthcare Service System Using Lifelog Based Context Computing

Intelligent bio-sensor information processing was developed using lifelog based context aware technology to provide a flexible and dynamic range of diagnostic capabilities to satisfy healthcare requirements in ubiquitous and mobile computing environments. To accomplish this, various noise signals were grouped into six categories by context estimation and effectively reconfigured noise reduction filters by neural network and genetic algorithm. The neural network-based control module effectively selected an optimal filter block by noise context-based clustering in running mode, and filtering performance was improved by genetic algorithm in evolution mode. Due to its adaptive criteria, genetic algorithm was used to explore the action configuration for each identified bio-context to implement our concept. Our proposed Bio-interactive healthcare service system adopts the concepts of biological context-awareness with evolutionary computations in working environments modeled and identified as bio-sensors based environmental contexts. We used an unsupervised learning algorithm for lifelog based context modeling and a supervised learning algorithm for context identification.

Real-Time Tracking and Recognition Systems for Interactive Telemedicine Health Services

Recent changes affecting the health industry include the digitization of medical information as well as the exchange of medical information through a network-connected medical infrastructure. In this paper, we propose a real-time tracking and recognition system for interactive telemedicine health services. The proposed method is a methodology for both hand and finger detection applied to posture recognition in telemedicine. The detected hand or finger can be used to implement a non-contact mouse in the machine-to-machine. This technology can be used to control telemedicine health devices such as a public healthcare system, pedometer health information reader, glucose-monitoring device, and blood pressure gauge. Skin color is used to segment the hand region from the background, and the contour is extracted from the segmented hand. Contour analysis provides the locations of the fingertips on the hand. Fingertip tracking is performed using a constant velocity model with a pixel-labeling approach. From the tracking process, several hand features can be extracted and then fed into a finite state classifier to identify the hand configuration. The hand can be classified into many gesture classes or several different movement directions. Using this method, we performed an extensive experiment and obtained a very encouraging result. It is shown that using the method used in previous studies, some of the points are lost, whereas using the proposed method described in this paper, all lost points are recovered with no or little displacement error. Ultimately, this paper provides empirical verification of the adequacy and validity of the proposed system for telemedicine health services. Accordingly, the satisfaction and quality of services will improve gesture recognition for interactive telemedicine health services.

Ontology-based inference system for adaptive object recognition

This paper presents a statistical ontology approach for adaptive object recognition in a situation-variant environment. We propose a context model based on statistical ontology that is concentrated on object recognition. Due to the effects of illumination on a supreme obstinate designing context-sensitive recognition system, we focused on designing a context-variant system using statistical ontology. Ontology, a collection of concepts and their interrelationships, provides an abstract view of an application domain. Researchers produce ontologies in order to understand and explain underlying principles and environmental factors. In this paper, we propose an ontology-based inference system for adaptive object recognition. The proposed method utilizes context ontology, context modeling, context adaptation, and context categorization to design the ontology based on illumination criteria for surveillance. After selecting the proper ontology domain, a set of actions is selected that produces better performance in that domain. We also carried out extensive experiments on these concepts in the area of object recognition in a dynamic changing environment, achieving enormous success that will enable us to proceed with our basic concepts.

Development of Real-Time Gesture Recognition System Using Visual Interaction

The aim of this paper is to present a methodology for hand detection, propose a finger detection method, and finally apply them to posture recognition. The detected hand and finger can be used to implement the non-contact mouse. This technology can be used to control home devices such as curtain and television. Skin color is used to segment the hand region from the background, and counter is extracted from the segmented hand. The counter analysis of gives us the location of fingertip in the hand. Fingertip tracking is performed assuming a constant velocity model and using a pixel labeling approach. From the tracking process, we extract several hand features that are fed to a finite state classifier that identifies the hand configuration. The hand can be classified into many gesture classes or several different movement directions. This method of skin segmentation assumes that the background does not contain any skin colored object beside hands. We have performed an extensive experiment and achieved a very encouraging result. Ultimately, this paper suggests an empirical application to verify the adequacy and validity of the proposed systems. Accordingly, the satisfaction and quality of services will improved gesture recognition.

Skin Color Based Hand and Finger Detection for Gesture Recognition in CCTV Surveillance

In this paper, we proposed the skin color based hand and finger detection technology for the gesture recognition in CCTV surveillance. The aim of this paper is to present the methodology for hand detection and propose the finger detection method. The detected hand and finger can be used to implement the non-contact mouse. This technology can be used to control the home devices such as home-theater and television. Skin color is used to segment the hand region from background and contour is extracted from the segmented hand. Analysis of contour gives us the location of finger tip in the hand. After detecting the location of the fingertip, this system tracks the fingertip by using only R channel alone, and in recognition of hand motions to apply differential image, such as the removal of useless image shows a robust side. We explain about experiment which relates in fingertip tracking and finger gestures recognition, and experiment result shows the accuracy above 96%.

서베일런스에서 피셔의 선형 판별 분석을 이용한 사람 검출의 성능 향상

Many reported methods assume that the people in an image or an image sequence have been identified and localization. People detection is one of very important variable to affect for the systems performance as the basis technology about the detection of other objects and interacting with people and computers, motion recognition. In this paper, we present an efficient linear discriminant for multi-view people detection. Our approaches are based on linear discriminant. We define training data with fisher Linear discriminant to efficient learning method. People detection is considerably difficult because it will be influenced by poses of people and changes in illumination. This idea can solve the multi-view scale and people detection problem quickly and efficiently, which fits for detecting people automatically. In this paper, we extract people using fisher linear discriminant that is hierarchical models invariant pose and background. We estimation the pose in detected people. The purpose of this paper is to classify people and non-people using fisher linear discriminant.

Context-Aware Statistical Inference System for Effective Object Recognition.

This paper proposes a statistical ontology approach for adaptive object recognition in a situation-variant environment. In this paper, we introduce a new concept, statistical ontology, for context sensitivity, as we found that many developed systems work in a context-invariant environment. Due to the effects of illumination on a supreme obstinate designing context-sensitive recognition system, we have focused on designing such a context-variant system using statistical ontology. Ontology can be defined as an explicit specification of conceptualization of a domain typically captured in an abstract model of how people think about things in the domain. People produce ontologies to understand and explain underlying principles and environmental factors. In this research, we have proposed context ontology, context modeling, context adaptation, and context categorization to design ontology based on illumination criteria. After selecting the proper ontology domain, we benefit from selecting a set of actions that produces better performance on that domain. We have carried out extensive experiments on these concepts in the area of object recognition in a dynamic changing environment, and we have achieved enormous success, which will enable us to proceed on our basic concepts.

Evolutionary Bio-Interaction Knowledge Accumulation for Smart Healthcare

The range of ubiquitous computing technology available for use in healthcare continues to evolve, allowing for an increasing variety of wireless sensors, devices, and actuators to be deployed in changing environments. This paper presents a robust distributed architecture for adaptive and intelligent bio-interaction systems, called Evolutionary Bio-inspired Knowledge Accumulation. This system is designed to its capability to increase knowledge enhancement even in dynamic and uneven environments. Our proposed system adopts the concepts of biological context-awareness with evolutionary computations where the working environments are modeled and identified as bio-environmental contexts. We have used an unsupervised learning algorithm for bio-context modeling, and a supervised learning algorithm for context identification. A genetic algorithm, for its adaptive criteria, is used to explore action configuration for each identified bio context to implement our concept. This framework has been used to reduce noise in ECG signals that have been gathered in routine remote healthcare monitoring. Experimental results showed that the proposed algorithm effectively removes baseline wander noise and muscle noise, and feature extraction results showed a significant improvement of T duration extraction values.

Color Area Correction Algorithm for Tracking Curved Fingertip

In the field of image processing to track the fingertip much research has been done. The most common way to calculate the fingertip first, to extract color information. Then, it uses Blob Coloring algorithms which are expressed in blob functions the skin contour and calculates. The algorithm from contour decides the highest location with the fingertip. But this method when measuring it location from the finger condition which bents is not the actual fingertip and has the problem which detects the location which goes wrong. This paper proposes the color space correction algorithm to tracks the fingertip which bents. The method which proposes when tracking the fingertip from the finger condition which bents solves the problem which measures the location which goes wrong. Aim of this paper in compliance with the propensity of the users forecasts a problem in advance and corrects with improvement at the time of height boil an efficiency. Ultimately, this paper suggests empirical application to verify the adequacy and the validity with the proposed method. Accordingly, the satisfaction and the quality of services will be improved the image recognition.