• Korean Journal of Remote Sensing
• /
• v.36 no.5_2
• /
• pp.949-959
• /
• 2020

In this paper, as a prior study related to the generation of topographic information using the CAS500-1/2 satellite, we propose a method of extraction DSM for each 1:5000 scaled map in North Korea using KOMPSAT-3A strip images. This technique is designed to set the processing area by receiving shape file, only to generate output for every 1:5000 scaled map. In addition, dense point clouds and the DSM were extracted by applying MDR, a robust stereo image matching technique. Considering that the strip images are input in the units of scenes, we attempted to extract a DSM by processing and merging multiple image pairs in one 1:5000 map area. As a result, it was possible to confirm the generation of an integrated DSM with minimal separation at the junction, and as a result of the accuracy analysis, it was confirmed that the accuracy was within 5m compared to GCP.

• The Journal of the Acoustical Society of Korea
• /
• v.14 no.2E
• /
• pp.91-97
• /
• 1995

This paper describes the implementation of a multi-channel audio codec for HETV. This codec has the features of the 3/2-stereo plus low frequency enhancement, downward compatibility with the smaller number of channels, backward compatibility with the existing 2/0-stereo system(MPEG-1 audio), and multilingual capability. The encoder of this codec consists of 6-channel analog audio input part with the sampling rate of 48 kHz, 4-channel digital audio input part and three TMS320C40 /DSPs. The encoder implements multi-channel audio compression using a human perceptual psychoacoustic model, and has the bit rate reduction to 384 kbit/s without impairment of subjective quality. The decoder consists of 6-channel analog audio output part, 4-channel digital audio output part, and two TMS320C40 DSPs for a decoding procedure. The decoder analyzes the bit stream received with bit rate of 384 kbit/s from the encoder and reproduces the multi-channel audio signals for analog and digital outputs. The multi-processing of this audio codec using multiple DSPs is ensured by high speed transfer of date between DSPs through coordinating communication port activities with DMA coprocessors. Finally, some technical considerations are suggested to realize the problem of real-time operation, which are found out through the implementation of this codec using the MPEG-2 layer II sudio coding algorithm and the use of the hardware architecture with commercial multiple DSPs.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.8
• /
• pp.341-354
• /
• 2023

The title of a document is the brief summarization of the document. Readers can easily understand a document if we provide them with its title in their preferred styles and the languages. In this research, we propose a cross-lingual and style-based title generation model using multiple adapters. To train the model, we need a parallel corpus in several languages with different styles. It is quite difficult to construct this kind of parallel corpus; however, a monolingual title generation corpus of the same style can be built easily. Therefore, we apply a zero-shot strategy to generate a title in a different language and with a different style for an input document. A baseline model is Transformer consisting of an encoder and a decoder, pre-trained by several languages. The model is then equipped with multiple adapters for translation, languages, and styles. After the model learns a translation task from parallel corpus, it learns a title generation task from monolingual title generation corpus. When training the model with a task, we only activate an adapter that corresponds to the task. When generating a cross-lingual and style-based title, we only activate adapters that correspond to a target language and a target style. An experimental result shows that our proposed model is only as good as a pipeline model that first translates into a target language and then generates a title. There have been significant changes in natural language generation due to the emergence of large-scale language models. However, research to improve the performance of natural language generation using limited resources and limited data needs to continue. In this regard, this study seeks to explore the significance of such research.

• Journal of the Institute of Convergence Signal Processing
• /
• v.24 no.2
• /
• pp.103-111
• /
• 2023

Consensus, that aims to converge the states of agents to the same states through information exchanges between agents, has been widely studied to control the multi-agent systems. In real systems, the measurement variables of each agent may be different, the loss of information across communication may occur, and the different networks for each state may need to be constructed for safety. Moreover, the input saturation and the disturbances in the system may cause instability. Therefore, this paper studies the PID(Proportional-Integral-Derivative)-based consensus control to achieve the swarm behavior of the multi-agent systems considering the heterogeneous state information, the input saturations, and the disturbances. Specifically, we consider the multiple follower agents and the single leader agent modeled by the second-order systems, and investigate the conditions to achieve the consensus based on the stability of the error system. It is confirmed that the proposed algorithm can achieve the consensus if only the connectivity of the position graph is guaranteed. Moreover, by extending the consensus algorithm, we study the formation control problem for the multi-agent systems. Finally, the validity of the proposed algorithm was verified through the simulations.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.8B
• /
• pp.691-700
• /
• 2006

Based on the destination IP address of incoming packets, the Internet routers determine next hops and forward packets toward final destinations through If address lookup. The bandwidth of communication links increases exponentially fast as well as the routing table size grows significant as the number of single host networks attached to the Internet increases. Since packets should be processed at wire-speed, the increased link speed reduces the processing time of a packet in routers, and hence more efficient and fast IP address lookup algorithms and architectures are required in the next generation routers. Most of the previous IP lookup schemes compare routing prefixes of shorter length first with a given input IP address. Since IP address lookup needs to find the most specific route of the given input, search continues until the longest matched prefix is found while it keeps remembering the current test matching prefix. In this paper, based on binary search on prefix length, we proposed a new IP address lookup algorithm which compares longer prefixes first. The proposed scheme is consisted of multiple tries with prefixes on leaves only. The trie composed of the longest prefixes is primarily searched whether there is a match with the given input. This processing is repeated for the trio of the next longer prefixes until there finds a match. Hence the proposed algorithm provides the fast search speed. The proposed algorithm also provides the incremental update of prefixes while the previous binary search on length scheme does not provide the incremental update because of pre-processing requirement. In this paper, we performed extensive simulations and showed the performance comparisons with related works.

• Korean Chemical Engineering Research
• /
• v.46 no.3
• /
• pp.486-491
• /
• 2008

This paper describes a model identification method that has been applied to a commercial 12-inch RTP (rapid thermal processing) equipment with an ultimate aim to develop a high-performance advanced controller. Seven thermocouples are attached on the wafer surface and twelve tungsten-halogen lamp groups are used to heat up the wafer. To obtain a MIMO balanced state space model, multiple SIMO (single-input multiple-output) identification with highorder ARX models have been conducted and the resulting models have been combined, transformed and reduced to a MIMO balanced state space model through a balanced truncation technique. The identification experiments were designed to minimize the wafer warpage and an output linearization block has been proposed for compensation of the nonlinearity from the radiation-dominant heat transfer. As a result from the identification at around 600, 700, and $800^{\circ}C$, respectively, it was found that $y=T(K)^2$ and the state dimension of 80-100 are most desirable. With this choice the root-mean-square value of the one-step-ahead temperature prediction error was found to be in the range of 0.125-0.135 K.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.67 no.1
• /
• pp.114-123
• /
• 2018

This paper concerns a design and learning method of softmax function neural networks based on K-means clustering. The partial discharge data Information is preliminarily processed through simulation using an Epoxy Mica Coupling sensor and an internal Phase Resolved Partial Discharge Analysis algorithm. The obtained information is processed according to the characteristics of the pattern using a Motor Insulation Monitoring System program. At this time, the processed data are total 4 types that void discharge, corona discharge, surface discharge and slot discharge. The partial discharge data with high dimensional input variables are secondarily processed by principal component analysis method and reduced with keeping the characteristics of pattern as low dimensional input variables. And therefore, the pattern classifier processing speed exhibits improved effects. In addition, in the process of extracting the partial discharge data through the MIMS program, the magnitude of amplitude is divided into the maximum value and the average value, and two pattern characteristics are set and compared and analyzed. In the first half of the proposed partial discharge pattern classifier, the input and hidden layers are classified by using the K-means clustering method and the output of the hidden layer is obtained. In the latter part, the cross entropy error function is used for parameter learning between the hidden layer and the output layer. The final output layer is output as a normalized probability value between 0 and 1 using the softmax function. The advantage of using the softmax function is that it allows access and application of multiple class problems and stochastic interpretation. First of all, there is an advantage that one output value affects the remaining output value and its accompanying learning is accelerated. Also, to solve the overfitting problem, L2-normalization is applied. To prove the superiority of the proposed pattern classifier, we compare and analyze the classification rate with conventional radial basis function neural networks.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.47 no.4
• /
• pp.41-49
• /
• 2010

This paper proposes a new high speed SHA-1 architecture using multiple unfolding and pre-computation techniques. We unfolds iterative hash operations to 2 continuos hash stage and reschedules computation timing. Then, the part of critical path is computed at the previous hash operation round and the rest is performed in the present round. These techniques reduce 3 additions to 2 additions on the critical path. It makes the maximum clock frequency of 118 MHz which provides throughput rate of 5.9 Gbps. The proposed architecture shows 26% higher throughput with a 32% smaller hardware size compared to other counterparts. This paper also introduces a analytical model of multiple SHA-1 architecture at the system level that maps a large input data on SHA-1 block in parallel. The model gives us the required number of SHA-1 blocks for a large multimedia data processing that it helps to make decision hardware configuration. The hs fospeed SHA-1 is useful to generate a condensed message and may strengthen the security of mobile communication and internet service.

• Journal of KIISE:Software and Applications
• /
• v.32 no.6
• /
• pp.554-569
• /
• 2005

In the semistructured distributed documents, it is very difficult to formalize and implement the query processing system due to the lack of structure and rule of the data. In order to precisely retrieve and process the heterogeneous semistructured documents, it is required to handle multiple mappings such as 1:1, 1:W and W:1 on an element simultaneously and to generate the schema from the distributed documents. In this paper, we have proposed an query processing algorithm for querying and answering on the heterogeneous semistructured data or documents over distributed systems and implemented with a metadata interface. The algorithm for generating local queries from the global query consists of mapping between g1oba1 and local nodes, data transformation according to the mapping types, path substitution, and resolving the heterogeneity among nodes on a global input query with metadata information. The mapping, transformation, and path substitution algorithms between the global schema and the local schemas have been implemented the metadata interface called DBXMI (for Distributed Documents XML Metadata Interface). The nodes with the same node name and different mapping or meanings is resolved by automatically extracting node identification information from the local schema automatically. The system uses Quilt as its XML query language. An experiment testing is reported over 3 different OEM model semistructured restaurant documents. The prototype system is developed under Windows system with Java and JavaCC compiler.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.6
• /
• pp.84-90
• /
• 2020

As the uploading and downloading of data through the Internet is becoming more common, data including personal information are easily exposed to unauthorized users. In this study, we detect a target area in images that contain personal information, except for the background, and we protect the detected target area by using a blocking method suitable for the surrounding situation. In this method, only the target area from color image input containing personal information is segmented based on skin color. Subsequently, blurring of the corresponding area is performed in multiple stages based on the surrounding situation to effectively block the detected area, thereby protecting the personal information from being exposed. Experimental results show that the proposed method blocks the object region containing personal information 2.3% more accurately than an existing method. The proposed method is expected to be utilized in fields related to image processing, such as video security, target surveillance, and object covering.