Mi Young Nam

Color Based Hand and Finger Detection Technology for User Interaction

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 curtain and television. Skin color is used to segment the hand region from background and counter is extracted from the segmented hand. Analysis of counter gives us the location of finger tip in the hand. We have performed extensive experiment and achieve very encouraging result.

Multi-class Multi-object Tracking Using Changing Point Detection

This paper presents a robust multi-class multi-object tracking (MCMOT) formulated by a Bayesian filtering framework. Multi-object tracking for unlimited object classes is conducted by combining detection responses and changing point detection (CPD) algorithm. The CPD model is used to observe abrupt or abnormal changes due to a drift and an occlusion based spatiotemporal characteristics of track states. The ensemble of convolutional neural network (CNN) based object detector and Lucas-Kanede Tracker (KLT) based motion detector is employed to compute the likelihoods of foreground regions as the detection responses of different object classes. Extensive experiments are performed using lately introduced challenging benchmark videos; ImageNet VID and MOT benchmark dataset. The comparison to state-of-the-art video tracking techniques shows very encouraging results.

Adaptive feature representation for robust face recognition using context-aware approach

In this paper, we investigate how to preprocess and accurately identify features from both normal and bad input images, for robust face recognition under uneven illumination environments. Bad illumination is the most challenging problem when implementing robust face recognition systems. From extensive experiments, we found that the performance of individual filtering methods for image enhancement is highly dependent upon application environments. For example, retinex provides good performance under bad illumination. However, it provides very poor performance under normal illumination. On the other hand, histogram equalization provides sufficiently good performance for normal images, however, performance falls dramatically under bad illumination. Since no prior knowledge of the system-working environment can be assumed, the proposed method tries to provide adaptive preprocessing as well as feature representation configuration by exploring the filter combination based on illumination context-awareness. The proposed method has been tested on Inha and FERET image datasets at the preprocessing and feature representation stages for robust face recognition under uneven illumination. Extensive experiments show the proposed system can achieve exceptional performance in varying illumination environments.

Context-Aware evolvable system framework for environment identifying systems

This paper proposes a novel framework for adaptive and intelligent systems that can be used under dynamic and uneven environments by taking advantage of environment context identification. Adaptation to dynamically changing environments is very important since advanced applications become pervasive and ubiquitous. The proposed framework, callesd CAES (Context-Aware Evolvable System), adopts the concept of context-aware and the evolutionary computing, and the system working environments are learned (clustered) and identified as environmental contexts. The context-awareness has been carried out by unsupervised learning, Fuzzy ART. Genetic algorithm (GA) is used to explore the most effective action configuration for each identified context. The knowledge of the individual context and its associated chromosomes representing optimal action configurations is accumulated and stored in the context knowledge base. Once the context knowledge is constructed, the system can adapt to varying environment in real-time. The framework of CAES has been tested in the area of intelligent vision application where most approaches show vulnerability under dynamically changing environments. The superiority of the proposed scheme is shown using three face image data sets: Inha, FERET, and Yale.

Pupil location and movement measurement for efficient emotional sensibility analysis

With advances in intelligent technologies, e.g. ambient intelligence, context-aware, and pervasive systems, much research is now devoted to a computational paradigm that senses and perceives changes in human emotion. This paper presents a context-aware architecture for adaptive emotional sensibility analysis called CAF-ESA (a Context-Aware Framework based Emotional Sensibility Analysis) with adaptive capability for use under both diverse changes in both human emotion and the illumination environment. Our proposed system implements context-awareness by a system that identifies working situations as usage contexts. An unsupervised learning algorithm models usage context while a supervised learning algorithm identifies the usage context. A genetic algorithm explores the emotional sensibility space for each identified usage context to determine human eye movement face images. The framework is validated for locating the pupil under changing illumination environments, and for pupil movement that is associated with emotional sensibility such as for both a positive and a negative emotion. We have achieved encouraging experimental results in the real time detection of pupil location and measurement of pupil movement.

Facial landmark detection system using interest-region model and edge energy function

In this paper, we proposed a new facial landmark-detection system using as edge energy function. The facial landmark-detection system is divided into a learning stage and a detection stage. The learning stage creates an interest-region model, to set up a search region of each landmark, as pre-information necessary for a detection stage and creates a detector for each landmark to detect a landmark in a search region. The detection stage sets up a search region of each landmark in an input image with an interest-region model created in the learning stage. Because a landmark to detect from a system has the characteristics of an edge as both edge of an eye, both edge of a mouth and both edges of eyebrows, we have detected a landmark by applying an edge energy function to the Bayesian discrimination method. We have implemented aforementioned technique by abstracting 800 impassive images from the FERET database and have measured data in which the normalized average error distance is less than 0.1 occupying 98 % of the total data.

An efficient face location using integrated feature space

We propose a method for an efficient frontal face detection using skin color, integrated feature space, and post processing. The proposed method reduces the search space by facial color information and detects face candidate windows by integrated feature space. The integrated feature space consists of intensity and texture information. Multiple Bayesian classifiers are employed for selection of face candidate windows on integrated feature space. And we use face and face-like nonface samples to training these Bayesian classifiers. Finally, face regions of the detected candidates are selected by merging and filtering post processing.

Human face detection using skin color context awareness and context-based bayesian classifiers

We propose a cascade detection scheme by combining the color feature-based method and appearance-based method. In addition, the scheme employs illumination context-awareness so that the detection scheme can react in a robust way against dynamically changing illumination. Skin color provides rich information for extracting rough area of the face. Difficulties in detecting face skin color come from the variations in ambient light, image capturing devices, etc,. Appearance-based object detection, multiple Bayesian classifiers here, is attractive since it could accumulate object models by autonomous learning process. This approach can be easily adopted in searching for multiple scale faces by scaling up/down the input image with some factor. The appearance-based method shows more stability under changing illumination than other detection methods, but it is still bordered from the variations in illumination. We employ Fuzzy ART and RBFN for the illumination context- awareness. The proposed face detection achieves the capacity of the high level attentive process by taking advantage of the illumination context-awareness in both color feature-based detection and multiple Bayesian classifiers. We achieve very encouraging experimental results, especially when illumination condition varies dynamically.

Adaptive gabor wavelet for efficient object recognition

This paper describes, using situational awareness and Genetic algorithm, a run-time optimization methodology of the Gabor wavelet parameters so that it produces a feature space for efficient object recognition. Gabor wavelet efficiently extracts the feature space of orientation selectivity, spatial frequency and spatial localization. Most previous object recognition approaches using Gabor wavelet do not include systematic optimization of the parameters for the Gabor kernel, even though the system performance might be much sensitive to the characteristics of the Gabor parameters. This paper explores efficient object recognition using adaptive Gabor wavelet based situational aware method. The superiority of the proposed system is shown using IT-Lab, FERET and Yale face database. We achieved encouraging experimental results.

Adaptive classifier combination for visual information processing using data context-awareness

This paper addresses a novel method of classifier combination for efficient object recognition using data context-awareness called “Adaptable Classifier Combination (ACC)”. The proposed method tries to distinguish the context category of input image data and decides the classifier combination structure accordingly by Genetic algorithm. It stores its experiences in terms of the data context category and the evolved artificial chromosome so that the evolutionary knowledge can be used later. The proposed method has been evaluated in the area of face recognition. Most previous face recognition schemes define their system structures at the design phases, and the structures are not adaptive during operation. Such approaches usually show vulnerability under varying illumination environment. Data context-awareness, modeling and identification of input data as data context categories, is carried out using SOM(Self Organized Map). The face data context are described based on the image attributes of light direction and brightness. The proposed scheme can adapt itself to an input data in real-time by identifying the data context category and previously derived chromosome. The superiority of the proposed system is shown using four data sets: Inha, FERET and Yale DB.