• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.10C
• /
• pp.983-991
• /
• 2005

Human faces are commonly found in a video such as a drama and provide useful information for video content analysis. Therefore, face detection plays an important role in applications such as face recognition, and face image database management. In this paper, we propose a face detection algorithm based on pre-processing of scene change detection for indexing and cast searching in video. The proposed algorithm consists of three stages: scene change detection stage, face region detection stage, and eyes and mouth detection stage. Experimental results show that the proposed algorithm can detect faces successfully over a wide range of facial variations in scale, rotation, pose, and position, and the performance is improved by $24\%$with profile images comparing with conventional methods using color components.

• Journal of Korea Multimedia Society
• /
• v.21 no.8
• /
• pp.888-896
• /
• 2018

In this paper, we report detection and recognition of vehicle logo from images captured from street CCTV. Image data includes both the front and rear view of the vehicles. The proposed method is a two-step process which combines image preprocessing and faster region-based convolutional neural network (R-CNN) for logo recognition. Without preprocessing, faster R-CNN accuracy is high only if the image quality is good. The proposed system is focusing on street CCTV camera where image quality is different from a front facing camera. Using perspective transformation the top view images are transformed into front view images. In this system, the detection and accuracy are much higher as compared to the existing algorithm. As a result of the experiment, on day data the detection and recognition rate is improved by 2% and night data, detection rate improved by 14%.

• Journal of Institute of Control, Robotics and Systems
• /
• v.16 no.9
• /
• pp.827-832
• /
• 2010

Many researchers have tried to detect the falling and to reduce the injury associated with falling. Normally the method of detection of a loss of balance is more efficient than that of a compensatory motion in order to predict the falling. The detection algorithm of the loss of balance was composed of three main parts: parts of processing of measured data, construction of an internal model and detection of the loss of balance. The internal model represented a simple dynamic motion balancing with two rear legs of a four-legged chair and was a simplified model of a central nervous system of a person. The internal model was defined by the experimental data obtained within a fixed time interval, and was applied to the detecting algorithm to the end of the experiment without being changed. The balancing motion controlled by the human brain was improved in process of time because of the experience accruing to the brain from controlling sensory organs. In this study a reconstruction method of the internal model was used in order to improve the success rate and the detecting time of the algorithm and was changed with time the same as the brain did. When using the reconstruction method, the success rate and the detecting time were 95 % and 0.729 sec, respectively and those results were improved by about 7.6 % and 0.25 sec in comparison to the results of the paper of Ahmed and Ashton-Miller. The results showed that the proposed reconstruction method of the internal model was efficient to improve the detecting performance of the algorithm.

• The KIPS Transactions:PartB
• /
• v.9B no.4
• /
• pp.491-500
• /
• 2002

In this paper, we propose a real-time face detection and recognition system by using skin color informations, geometrical feature vectors of face, and facial angle informations from color face image. The proposed algorithm improved face region extraction efficiency by using skin color informations on the HSI color coordinate and face edge information. And also, it improved face recognition efficiency by using geometrical feature vectors of face and facial angles from the extracted face region image. In the experiment, the proposed algorithm shows more improved recognition efficiency as well as face region extraction efficiency than conventional methods.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.410-412
• /
• 2021

This paper proposes an improved pedestrian detection method using object and background histograms. Objects detected through the HOG & SVM algorithm are detected in a square shape. Inside the square area, the background and the object area are mixed. If only the area of the object excluding the background is detected, various object-related information may be easily obtained. The size of the detected rectangle is readjusted using an xy-axis projection algorithm to fit the size of the object. And then, the improved object is detected by dividing the background and the object based on the histogram of the object in the readjusted square. The average values of precision and recall, which are reliability evaluations comparing the detected object with the original object, are 97.9% and 90%, respectively.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.7
• /
• pp.3076-3092
• /
• 2020

An IKPCA-ELM-based intrusion detection method is developed to address the problem of the low accuracy and slow speed of intrusion detection caused by redundancies and high dimensions of data in the network. First, in order to reduce the effects of uneven sample distribution and sample attribute differences on the extraction of KPCA features, the sample attribute mean and mean square error are introduced into the Gaussian radial basis function and polynomial kernel function respectively, and the two improved kernel functions are combined to construct a hybrid kernel function. Second, an improved particle swarm optimization (IPSO) algorithm is proposed to determine the optimal hybrid kernel function for improved kernel principal component analysis (IKPCA). Finally, IKPCA is conducted to complete feature extraction, and an extreme learning machine (ELM) is applied to classify common attack type detection. The experimental results demonstrate the effectiveness of the constructed hybrid kernel function. Compared with other intrusion detection methods, IKPCA-ELM not only ensures high accuracy rates, but also reduces the detection time and false alarm rate, especially reducing the false alarm rate of small sample attacks.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.40 no.9
• /
• pp.1828-1836
• /
• 2015

Object tracking algorithms can reduce computational cost by avoiding computation over the whole image through the selection of region of interests based on object detection. So, accurate object detection is an important task for object tracking. The background subtraction algorithm has been widely used in moving object detection using a stationary camera. However, it has the problem of object detection error due to incorrect background modeling, whereas the method of background modeling has been improved by many researches. This paper proposes a new moving object detection algorithm to overcome the drawback of the conventional background subtraction algorithm by combining the background subtraction algorithm with the motion history image algorithm that is usually used in gesture detection. Although the proposed algorithm demands more processing time because of time taken for combining two algorithms, it meet the real-time processing requirement. Moreover, experimental results show that it has higher accuracy compared with the previous two algorithms.

• Current Optics and Photonics
• /
• v.2 no.1
• /
• pp.7-14
• /
• 2018

As a sophisticated detection technology, LIDAR has been widely employed to probe low-altitude windshear. Due to the drawbacks of the traditional ramp algorithm, the alarm accuracy of the LIDAR has not been satisfactory. Aiming at settling this matter, a novel method is proposed on the basis of improved signal smoothing and second windshear detection, which essentially acts as a combination of ramp algorithm and segmentation approach, involving the human factor as well as signal fluctuations. Experiments on the real and artificial signals verify our approach.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.34S no.10
• /
• pp.26-32
• /
• 1997

The economical detection of dulal-tone multifrequency(DTMF) signals is an important factor when developing cost-effective telecommunication equipment. Each channel have independently a DTMF receiver, and it informs the detected signal to processors of the TDX. This paper proposes the linear prediction algorithm for the spectrum analysis. As a experimental resutls, it can obtain the improved performance to the DTMF receivers and reduce the real-time processing and memory waste.

• Journal of Electrical Engineering and Technology
• /
• v.12 no.5
• /
• pp.2067-2078
• /
• 2017

In this paper, a new face tracking algorithm is proposed. The CamShift (Continuously adaptive mean SHIFT) algorithm shows unstable tracking when there exist objects with similar color to that of face in the background. This drawback of the CamShift is resolved by the proposed algorithm using Kinect's pixel-by-pixel depth information and the skin detection method to extract candidate skin regions in HSV color space. Additionally, even when the target face is disappeared, or occluded, the proposed algorithm makes it robust to this occlusion by the feature point matching. Through experimental results, it is shown that the proposed algorithm is superior in tracking performance to that of existing TLD (Tracking-Learning-Detection) algorithm, and offers faster processing speed. Also, it overcomes all the existing shortfalls of CamShift with almost comparable processing time.