• Proceedings of the Korea Contents Association Conference
• /
• 2009.05a
• /
• pp.889-894
• /
• 2009

Accompanied by the rapid development of Computer Vision, Visual surveillance has achieved great evolution with more and more complicated processing. However there are still many problems to be resolved for robust and reliable visual surveillance, and the cast shadow occurring in motion detection process is one of them. Shadow pixels are often misclassified as object pixels so that they cause errors in localization, segmentation, tracking and classification of objects. This paper proposes a novel cast shadow removal method. As opposed to previous conventional methods, which considers pixel properties like intensity properties, color distortion, HSV color system, and etc., the proposed method utilizes observations about edge patterns in the shadow region in the current frame and the corresponding region in the background scene, and applies Laplacian edge detector to the blob regions in the current frame and the background scene. Then, the product of the outcomes of application determines whether the blob pixels in the foreground mask comes from object blob regions or shadow regions. The proposed method is simple but turns out practically very effective for Gaussian Mixture Model, which is verified through experiments.

• Proceedings of the KSPS conference
• /
• 2007.05a
• /
• pp.207-210
• /
• 2007

As the Internet is prevalent in our life, harmful contents have been increasing on the Internet, which has become a very serious problem. Among them, pornographic video is harmful as poison to our children. To prevent such an event, there are many filtering systems which are based on the keyword based methods or image based methods. The main purpose of this paper is to devise a system that classifies the pornographic videos based on the audio information. We use Mel-Cepstrum Modulation Energy (MCME) which is modulation energy calculated on the time trajectory of the Mel-Frequency cepstral coefficients (MFCC) and MFCC as the feature vector and Gaussian Mixture Model (GMM) as the classifier. With the experiments, the proposed system classified the 97.5% of pornographic data and 99.5% of non-pornographic data. We expect the proposed method can be used as a component of the more accurate classification system which uses video information and audio information simultaneously.

• Transactions on Electrical and Electronic Materials
• /
• v.13 no.1
• /
• pp.10-15
• /
• 2012

In conventional, skin detection methods using for skin color definitions is based on prior knowledge. By experimentation, the threshold value for dividing the background from the skin region is determined subjectively. A drawback of such techniques is that their performance is dependent on a threshold value which is estimated from repeated experiments. To overcome this, the present paper introduces a skin region detection method. This method uses a histogram approximation based on the mean shift algorithm. This proposed method applies the mean shift procedure to a histogram of a skin map of the input image. It is generated by comparing with the standard skin colors in the $C_bC_r$ color space. It divides the background from the skin region by selecting the maximum value according to the brightness level. As the histogram has the form of a discontinuous function. It is accumulated according to the brightness values of the pixels. It is then, approximated by a Gaussian mixture model (GMM) using the Bezier curve technique. Thus, the proposed method detects the skin region using the mean shift procedure to determine a maximum value. Rather than using a manually selected threshold value, as in existing techniques this becomes the dividing point. Experiments confirm that the new procedure effectively detects the skin region.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.17 no.10
• /
• pp.2419-2424
• /
• 2013

In a conventional MOG algorithm, a small threshold for background decision causes the background recognition delay in a periodic background and a large threshold makes it recognize passing objects as background in a stationary background. This paper proposes the improved MOG algorithm using adaptive threshold. The proposed algorithm estimates changes of weight in the dominant model of the MOG algorithm both in the short and long terms, classifies backgrounds into the stationary and periodic ones, and assigns proper thresholds to them. The simulation results show that the proposed algorithm decreases the maximum number of frame in background recognition delay from 137 to 4 in the periodic background keeping the equal performance with the conventional algorithm in the stationary background.

• ETRI Journal
• /
• v.30 no.1
• /
• pp.89-100
• /
• 2008

We consider the feature recombination technique in a multiband approach to speaker identification and verification. To overcome the ineffectiveness of conventional feature recombination in broadband noisy environments, we propose a new subband feature recombination which uses subband likelihoods and a subband reliable-feature selection technique with an adaptive noise model. In the decision step of speaker recognition, a few very low unreliable feature likelihood scores can cause a speaker recognition system to make an incorrect decision. To overcome this problem, reliable-feature selection adjusts the likelihood scores of an unreliable feature by comparison with those of an adaptive noise model, which is estimated by the maximum a posteriori adaptation technique using noise features directly obtained from noisy test speech. To evaluate the effectiveness of the proposed methods in noisy environments, we use the TIMIT database and the NTIMIT database, which is the corresponding telephone version of TIMIT database. The proposed subband feature recombination with subband reliable-feature selection achieves better performance than the conventional feature recombination system with reliable-feature selection.

• IEIE Transactions on Smart Processing and Computing
• /
• v.2 no.5
• /
• pp.277-281
• /
• 2013

Modern speaker verification systems based on support vector machines (SVMs) use Gaussian mixture model (GMM) supervectors as their input feature vectors, and the maximum a posteriori (MAP) adaptation is a conventional method for generating speaker-dependent GMMs by adapting a universal background model (UBM). MAP adaptation requires the appropriate amount of input utterance due to the number of model parameters to be estimated. On the other hand, with limited utterances, unreliable MAP adaptation can be performed, which causes adaptation noise even though the Bayesian priors used in the MAP adaptation smooth the movements between the UBM and speaker dependent GMMs. This paper proposes a sparse MAP adaptation method, which is known to perform well in the automatic speech recognition area. By introducing sparse MAP adaptation to the GMM-SVM-based speaker verification system, the adaptation noise can be mitigated effectively. The proposed method utilizes the L0 norm as a regularizer to induce sparsity. The experimental results on the TIMIT database showed that the sparse MAP-based GMM-SVM speaker verification system yields a 42.6% relative reduction in the equal error rate with few additional computations.

• Journal of Korea Multimedia Society
• /
• v.15 no.4
• /
• pp.449-471
• /
• 2012

In intelligent surveillance systems, it is required to robustly track multiple people. Most of the previous studies adopted a Gaussian mixture model (GMM) for discriminating the object from the background. However, it has a weakness that its performance is affected by illumination variations and shadow regions can be merged with the object. And when two foreground objects overlap, the GMM method cannot correctly discriminate the occluded regions. To overcome these problems, we propose a new method of tracking and identifying multiple people. The proposed research is novel in the following three ways compared to previous research: First, the illuminative variations and shadow regions are reduced by an illumination normalization based on the median and inverse filtering of the L*a*b* image. Second, the multiple occluded and overlapped people are tracked by combining the GMM in the still image and the Lucas-Kanade-Tomasi (LKT) method in successive images. Third, with the proposed human tracking and the existing face detection & recognition methods, the tracked multiple people are successfully identified. The experimental results show that the proposed method could track and recognize multiple people with accuracy.

• Journal of Advanced Navigation Technology
• /
• v.17 no.1
• /
• pp.90-97
• /
• 2013

Recently, intelligent transportation system based on image processing has been developed. In this paper, we propose a collision detection algorithm based on the analysis of motion vectors and the staying periods of vehicles in intersections. Objects in the region of interest are extracted from the subtraction image between background images based on Gaussian mixture model and input images. Collisions and traffic jams are detected by analysing measured motion vectors of vehicles and their staying periods in intersections. Experiments are performed on video sequences actually recoded at intersections. Correct detection rate and false alarm rate are 85.7% and 7.7%, respectively.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.13 no.3
• /
• pp.571-578
• /
• 2018

Recently, deep learning techniques have shown superior performance in various kinds of pattern recognition. However, there have been some arguments whether the DNN performs better than the conventional machine learning techniques when classification experiments are done using a small amount of training data. In this study, we compared the performance of the conventional GMM and SVM with DNN, a kind of deep learning techniques, in audio event detection. When tested on the same data, DNN has shown superior overall performance but SVM was better than DNN in segment-based F-score.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.24 no.10B
• /
• pp.1902-1911
• /
• 1999

In this paper we present a content-based image retrieval algorithm using the visual feature vectors which describe the spatial characteristics of objects. The proposed technique uses the Gaussian mixture model(GMM) to represent multi-colored objects and the expectation maximization(EM) algorithm is employed to estimate the maximum likelihood(ML) parameters of the model. After image segmentation is performed based on GMM, the shape and color features are extracted from each object using Fourier descriptors and color histograms, respectively. Image retrieval consists of two steps: first, the shape-based query is carried out to find the candidate images whose objects have the similar shapes with the query image and second, the color-based query is followed. The experimental results show that the proposed algorithm is effective in image retrieving by using the spatial and visual features of segmented objects.