• Journal of Institute of Control, Robotics and Systems
• /
• v.6 no.5
• /
• pp.396-403
• /
• 2000

A thermal imaging system is implemented for the measurement and the analysis of the thermal distribution of the target objects. The main part of the system is a thermal camera in which a focal plane array typed sensor is introduced. The sensor detects the mid-range infrared spectrum of target objects and then it outputs a generic video signal which should be processed to form a frame thermal image. Here, a digital signal processor(DSP) is applied for the high speed processing of the sensor signals. The DSP controls analog-to-digital converter, performs correction algorithms and outputs the frame thermal data to frame buffers. With the frame buffers can be generated a NTSC signal and transferred the frame data to personal computer(PC) for the analysis and a monitoring of the thermal scenes. By performing the signal processing functions in the DSP the overall system achieves a simple configuration. Several experimental results indicate the performance of the overall system.

• Journal of Information Processing Systems
• /
• v.10 no.3
• /
• pp.335-354
• /
• 2014

The accuracy of training-based activity recognition depends on the training procedure and the extent to which the training dataset comprehensively represents the activity and its varieties. Additionally, training incurs substantial cost and effort in the process of collecting training data. To address these limitations, we have developed a training-free activity recognition approach based on a fuzzy logic algorithm that utilizes a generic activity model and an associated activity semantic knowledge. The approach is validated through experimentation with real activity datasets. Results show that the fuzzy logic based algorithms exhibit comparable or better accuracy than other training-based approaches.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.22 no.7
• /
• pp.1515-1522
• /
• 1997

Several image coding algorithms have been developed for the telecommunication and multimedia systems with high image quality and high compression ratio. In order to achieve low entropy and distortion, the system should pay great cost of computation time and memory. In this paper, the uniform cubic lattice is chosen for Lattice Vector Quantization (LVQ) because of its generic simplicity. As a transform coding, the Discrete Wavelet Transform (DWT) is applied to the images because of its multiresolution property. The proposed algorithm is basically composed of the biorthogonal DWT and the uniform cubic LVQ. The multiresolution property of the DWT is actively used to optimize the entropy and the distortion on the basis of the distortion-rate function. The vector codebooks are also designed to be optimal at each subimage which is analyzed by the biorthogonal DWT. For compression efficiency, the vector codebook has different dimension depending on the variance of subimage. The simulation results show that the performance of the proposed coding mdthod is superior to the others in terms of the computation complexity and the PSNR in the range of entropy below 0.25 bpp.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2002.05a
• /
• pp.134-138
• /
• 2002

The human faces do not have distinct features unlike other general objects. In general the features of eyes, nose and mouth which are first recognized when human being see the face are defined. These features have different characteristics depending on different human face. In this paper, We propose a face recognition algorithm using the hidden Markov model(HMM). In the preprocessing stage, we find edges of a face using the locally adaptive threshold scheme and extract features based on generic knowledge of a face, then construct a database with extracted features. In training stage, we generate HMM parameters for each person by using the forward-backward algorithm. In the recognition stage, we apply probability values calculated by the HMM to input data. Then the input face is recognized by the euclidean distance of face feature vector and the cross-correlation between the input image and the database image. Computer simulation shows that the proposed HMM algorithm gives higher recognition rate compared with conventional face recognition algorithms.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.9
• /
• pp.97-106
• /
• 1996

In this paper, we propose a face recognition algorithm using the hidden markov model and neural networks (HMM-NN). In the preprocessing stage, we find edges of a face using the locally adaptive threshold (LAT) scheme and extract features based on generic knowledge of a face, then construct a database with extracted features. In the training stage, we generate HMM parameters for each person by using the forward-backward algorithm. In the recognition stage, we apply probability vlaues calculated by the HMM to subsequent neural networks (NN) as input data. Computer simulation shows that the proposed HMM-NN algorithm gives higher recognition rate compared with conventional face recognition algorithms.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.6 no.1
• /
• pp.49-58
• /
• 2003

Compared with generic GPS receiver, post-processing software GPS receiver has many advantages for high dynamic vehicle tracking. It has the advantage of the application of various tracking algorithms and aiding schemes. The post-processing system observes the carrier phase measurement data from the recorded GPS signals, detects and isolates the cycle slip. The observed carrier phase data and the raw data of the reference station are processed by carrier phase DGPS scheme. And the integer ambiguity resolution algorithm is used for resolving single frequency carrier phase ambiguity. The results of static and real flight test are presented and show that the proposed GPS translator processing system satisfies submeter accuracy.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.11a
• /
• pp.171-174
• /
• 2022

Weakly-supervised learning is a widely adopted approach in video anomaly detection whereby only video labels are utilized instead of expensive frame-level annotations. Since the success of multi-instance learning (MIL), almost all recent approaches are based on maximizing the margin between the set of abnormal video snippets and those of normal video snippets. In this work, we present a simple contrastive approach for weakly supervised video anomaly detection (WS-VAD) with aims to enhance the performance of existing models. The method is generic in nature and introduces a loss function to encourage attraction of output features from the same video class and repel those from different video classes. Experimental results demonstrate our method can be applied to existing algorithms to improve detection accuracy in public video anomaly dataset.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.39 no.7
• /
• pp.30-37
• /
• 2002

In this paper, we propose a new adaptive slot assignment algorithm called In-VDSA in order to guarantee the QoS(Quality of Service) of VBR(Variable Bit Rate) traffics in wireless ATM and maximize efficiency in use of wireless channels. In the proposed algorithm, the status of terminal buffers is encoded in signed number on the GFC(Generic Flow Control) field of an ATM cell header and piggybacked. And also, the number of slots to be assigned to the next frame is adjusted effectively, which is different to methods in the conventional slot assignment algorithms. As a result, we can guarantee QoS such as CLR(Cell Loss Rate) and cell delay and achieve the higher utilization of channels. The validity of the proposed algorithm has been justified in performance by analysis through simulation results using the BONeS tool and comparison with conventional methods.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.22 no.2
• /
• pp.179-188
• /
• 2012

Packing techniques, one of malicious code detection and analysis avoidance techniques, change code to reduce size and make analysts confused. Therefore, malwares have more time to spread out and it takes longer time to analyze them. Thus, these kind of unpacking techniques have been studied to deal with packed malicious code lately. Packed programs are unpacked during execution. When it is unpacked, the data inside of the packed program are changed. Because of these changes, the entropy value of packed program is changed. After unpacking, there will be no data changes; thus, the entropy value is not changed anymore. Therefore, packed programs could be unpacked finding the unpacking point using this characteristic regardless of packing algorithms. This paper suggests the generic unpacking mechanism using the method estimating the unpacking point through the variation of entropy values.

• International Journal of Control, Automation, and Systems
• /
• v.1 no.2
• /
• pp.194-202
• /
• 2003

In this paper, we introduce a category of Multi-FNN (Fuzzy-Neural Networks) models, analyze the underlying architectures and propose a comprehensive identification framework. The proposed Multi-FNNs dwell on a concept of fuzzy rule-based FNNs based on HCM clustering and evolutionary fuzzy granulation, and exploit linear inference being treated as a generic inference mechanism. By this nature, this FNN model is geared toward capturing relationships between information granules known as fuzzy sets. The form of the information granules themselves (in particular their distribution and a type of membership function) becomes an important design feature of the FNN model contributing to its structural as well as parametric optimization. The identification environment uses clustering techniques (Hard C - Means, HCM) and exploits genetic optimization as a vehicle of global optimization. The global optimization is augmented by more refined gradient-based learning mechanisms such as standard back-propagation. The HCM algorithm, whose role is to carry out preprocessing of the process data for system modeling, is utilized to determine the structure of Multi-FNNs. The detailed parameters of the Multi-FNN (such as apexes of membership functions, learning rates and momentum coefficients) are adjusted using genetic algorithms. An aggregate performance index with a weighting factor is proposed in order to achieve a sound balance between approximation and generalization (predictive) abilities of the model. To evaluate the performance of the proposed model, two numeric data sets are experimented with. One is the numerical data coming from a description of a certain nonlinear function and the other is NOx emission process data from a gas turbine power plant.