David K. Han

Acoustic and visual signal based context awareness system for mobile application

In this paper, an acoustic and visual signal based context awareness system is proposed for a mobile application. In particular multimodal system is designed that can sense and determine, in real-time, user contextual information, such as where the user is or what the user does, by processing acoustic and visual signals from the suitable sensors available in a mobile device. A variety of contextual information, such as babble sound in cafeteria, user¿s movement, and etc., can be recognized by the proposed acoustic and visual feature extraction and classification methods. We first describe the overall structure of the proposed system and then the algorithm for each module performing detection or classification of various contextual scenarios is presented. Representative experiments demonstrate the superiority of the proposed system while the actual implementation of the proposed scheme into mobile device such as a smart-phone confirms the effectiveness and realization of the proposed system.

Joint patch clustering-based dictionary learning for multimodal image fusion

A clustering-based dictionary learning is proposed for multimodal image fusion.Patches from different sources are clustered with their structural similarities.A compact dictionary is constructed by combining principal components of clusters.Sparse coefficients are estimated by a simultaneous orthogonal matching pursuit.The proposed method requires lower processing time with better fusion quality. Constructing a good dictionary is the key to a successful image fusion technique in sparsity-based models. An efficient dictionary learning method based on a joint patch clustering is proposed for multimodal image fusion. To construct an over-complete dictionary to ensure sufficient number of useful atoms for representing a fused image, which conveys image information from different sensor modalities, all patches from different source images are clustered together with their structural similarities. For constructing a compact but informative dictionary, only a few principal components that effectively describe each of joint patch clusters are selected and combined to form the over-complete dictionary. Finally, sparse coefficients are estimated by a simultaneous orthogonal matching pursuit algorithm to represent multimodal images with the common dictionary learned by the proposed method. The experimental results with various pairs of source images validate effectiveness of the proposed method for image fusion task.

Gesture recognition using depth-based hand tracking for contactless controller application

This paper proposes a gesture recognition system capable of providing a contactless controller via depth-based hand tracking. Main proposal is a hand tracking algorithm in depth image by calculating a hand weighted probability. An implementation of the proposed system into a contactless controller application demonstrated its effectiveness.

License Plate Detection Using Local Structure Patterns

We address the problem of license plate detection invideo surveillance systems. The Adaboost based approach,known for relative ease of implementation, makes use ofdiscriminative features such as edges or Haar-like features.In this paper, we propose a novel detection algorithm basedon local structure patterns for license plate detection. Theproposed algorithm includes post-processing methods toreduce false positive rate using positional and colorinformation of license plates. Experimental resultsdemonstrate effectiveness of the proposed methodcompared

Hidden Markov Model on a unit hypersphere space for gesture trajectory recognition

In this paper, a Mixture of von Mises-Fisher (MvMF) Probability Density Function (PDF) is incorporated into a Hidden Markov Model (HMM) in order to model spatio-temporal data in a unit-hypersphere space. The parameter estimation formulae for MvMF-HMM are derived in a closed form. As an application for the proposed MvMF-HMM, hands gesture trajectory recognition task is considered. Modeling gesture trajectory on a unit-hypersphere inherently removes bias from a subjects arm length or distance between a subject and camera. In experiments with public datasets, InteractPlay and UCF Kinect, the proposed MvMF-HMM showed superior recognition performance compared to current state-of-the-art techniques.

Rule-based trajectory segmentation for modeling hand motion trajectory

In this paper, we propose a simple but effective method of modeling hand gestures based on the angles and angular change rates of the hand trajectories. Each hand motion trajectory is composed of a unique series of straight and curved segments. In our Hidden Markov Model (HMM) implementation, these trajectories are modeled as a connected series of states analogous to the series of phonemes in speech recognition. The novelty of the work presented herein is that it provides an automated process of segmenting gesture trajectories based on a simple set of threshold values in the angular change measure. In order to represent the angular distribution of each separated state, the von Mises distribution is used. A likelihood based state segmentation was implemented in addition to the threshold based method to ensure that the gesture sets are segmented consistently. The proposed method can separate each angular state of the training data at the initialization step, thus providing a solution to mitigate the ambiguities on initializing the HMM. The effectiveness of the proposed method was demonstrated by the higher recognition rates in the experiments compared to the conventional methods.

Multiple vehicle tracking based on regional estimation in nighttime CCD images

In this paper, we develop an image based tracking algorithm of multiple vehicles focused to effective detection and segmentation of moving objects for tracking under poor environmental conditions. In particular, we propose a novel image-tracking algorithm aimed at being robust to occlusion, false alarms, missed detection, and partial or multiple detection of target objects, adverse conditions considered as important issues in Intelligent Transportation System (ITS). Upon applying the Retinex algorithm as preprocessing to reduce the illumination effects at nighttime images, we apply a two-step tracking procedure, performing regional search and track. A regional estimation is first achieved based on a gating using probability data association, to initiate the search. We then invoke an object-oriented multiprocessing for multiple vehicle tracking under poor conditions. Representative experimental results show that the proposed method is effective in CCD images.

Adaptive height-modified histogram equalization and chroma correction in YCbCr color space for fast backlight image compensation☆

Abstract Automatic exposure controls in commercially available cameras often encounter difficulties in capturing scenes with backlight luminance which dominates the entire image. An Adaptive Height-Modified Histogram Equalization (AHMHE) algorithm is proposed as a compensation technique for backlight images. It simultaneously enhances contrast in both the dark and the bright areas without creating regions of degraded local contrast. Moreover AHMHE is an adaptive algorithm: thus it requires minimal user input, and its reduced computational requirement makes it suitable for real-time application. In addition to AHMHE, a chroma correction technique was applied to chroma components in the YCbCr color space to produce more vivid color images. A series of subjective and index evaluations were conducted to measure the resultant image quality improvements by the AHMHE and the chroma correction algorithms.

Single image dehazing with image entropy and information fidelity

In this paper, we propose a new single image dehazing approach based on information fidelity and image entropy. The global atmospheric light is estimated by quadtree subdivision using transformed hazy images. Then, transmission is estimated by an objective function which is comprised of information fidelity and image entropy at non-overlapped sub-block regions. This is further refined by a Weighted Least Squares (WLS) optimization procedure to alleviate block artifacts. We compared performance of the proposed method with conventional methods to validate its effectiveness in an experiment.

Selective Background Adaptation Based Abnormal Acoustic Event Recognition for Audio Surveillance

In this paper, a method for abnormal acoustic event recognition in an audio surveillance system is presented. We propose a recognition scheme based on a hierarchical structure using a feature combination of Mel-Frequency Cepstral Coefficient (MFCC), timbre, and spectral statistics. A selective background adaptation is proposed for robust abnormal acoustic event recognition in real-world situations. For training, we use a database containing 9 abnormal events (scream, glass breaking, and etc.) and 6 background noise types collected under various surveillance situations. Gaussian Mixture Model (GMM) is considered for classifying the representative abnormal acoustic events and for selecting the background noise for adaptation. Effectiveness of the proposed method is demonstrated via representative experimental results.