• Journal of Advanced Information Technology and Convergence
• /
• v.10 no.2
• /
• pp.61-72
• /
• 2020

Recently, researches that are using deep learning technology in various fields are being conducted. The fields include geographic map processing. In this paper, I propose a method to infer where the map area included in the image is. The proposed method generates and learns images including a map, detects map areas from input images, extracts character strings belonging to those map areas, and converts the extracted character strings into coordinates through geocoding to infer the coordinates of the input image. Faster R-CNN was used for learning and map detection. In the experiment, the difference between the center coordinate of the map on the test image and the center coordinate of the detected map is calculated. The median value of the results of the experiment is 0.00158 for longitude and 0.00090 for latitude. In terms of distance, the difference is 141m in the east-west direction and 100m in the north-south direction.

• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.1-25
• /
• 2020

In this paper, we suggest an application system architecture which provides accurate, fast and efficient automatic gasometer reading function. The system captures gasometer image using mobile device camera, transmits the image to a cloud server on top of private LTE network, and analyzes the image to extract character information of device ID and gas usage amount by selective optical character recognition based on deep learning technology. In general, there are many types of character in an image and optical character recognition technology extracts all character information in an image. But some applications need to ignore non-of-interest types of character and only have to focus on some specific types of characters. For an example of the application, automatic gasometer reading system only need to extract device ID and gas usage amount character information from gasometer images to send bill to users. Non-of-interest character strings, such as device type, manufacturer, manufacturing date, specification and etc., are not valuable information to the application. Thus, the application have to analyze point of interest region and specific types of characters to extract valuable information only. We adopted CNN (Convolutional Neural Network) based object detection and CRNN (Convolutional Recurrent Neural Network) technology for selective optical character recognition which only analyze point of interest region for selective character information extraction. We build up 3 neural networks for the application system. The first is a convolutional neural network which detects point of interest region of gas usage amount and device ID information character strings, the second is another convolutional neural network which transforms spatial information of point of interest region to spatial sequential feature vectors, and the third is bi-directional long short term memory network which converts spatial sequential information to character strings using time-series analysis mapping from feature vectors to character strings. In this research, point of interest character strings are device ID and gas usage amount. Device ID consists of 12 arabic character strings and gas usage amount consists of 4 ~ 5 arabic character strings. All system components are implemented in Amazon Web Service Cloud with Intel Zeon E5-2686 v4 CPU and NVidia TESLA V100 GPU. The system architecture adopts master-lave processing structure for efficient and fast parallel processing coping with about 700,000 requests per day. Mobile device captures gasometer image and transmits to master process in AWS cloud. Master process runs on Intel Zeon CPU and pushes reading request from mobile device to an input queue with FIFO (First In First Out) structure. Slave process consists of 3 types of deep neural networks which conduct character recognition process and runs on NVidia GPU module. Slave process is always polling the input queue to get recognition request. If there are some requests from master process in the input queue, slave process converts the image in the input queue to device ID character string, gas usage amount character string and position information of the strings, returns the information to output queue, and switch to idle mode to poll the input queue. Master process gets final information form the output queue and delivers the information to the mobile device. We used total 27,120 gasometer images for training, validation and testing of 3 types of deep neural network. 22,985 images were used for training and validation, 4,135 images were used for testing. We randomly splitted 22,985 images with 8:2 ratio for training and validation respectively for each training epoch. 4,135 test image were categorized into 5 types (Normal, noise, reflex, scale and slant). Normal data is clean image data, noise means image with noise signal, relfex means image with light reflection in gasometer region, scale means images with small object size due to long-distance capturing and slant means images which is not horizontally flat. Final character string recognition accuracies for device ID and gas usage amount of normal data are 0.960 and 0.864 respectively.

• Journal of the Korean Mathematical Society
• /
• v.59 no.2
• /
• pp.421-438
• /
• 2022

We investigate how closed string mirror symmetry is related to homological mirror symmetry, under the presence of an explicit geometric mirror functor.

• East Asian mathematical journal
• /
• v.25 no.2
• /
• pp.215-220
• /
• 2009

In this paper, we investigate the generating strings and the number of 2-(edge)connected bipartite Steinhaus graphs.

• Journal of Korea Game Society
• /
• v.12 no.1
• /
• pp.57-65
• /
• 2012

Interactive games allow users to obtain embodied experience using the bodies as controllers. The same is true in interactive media arts where users engage in active participation. In contrast to video games, physical body feedback is desired and such practice can be found in robotic arts. I suggest that interactive media arts and interactive games should share common foundations. In this context, I introduce and explain an interactive robotic art work implemented. This work is a musical instrument that employs a robot which travels sitting on two strings in response to audience positions. In results, the robot modulates the vibrations of the strings by causing the effective lengths of the strings changed. The robot uses an economic multi-cell proximity sensor in order to track the audience. In the interaction, phenomenological tension could take place in the performative narrative space. In this paper, I discuss this interactive robotic work in the context of interactive games with a few examples.

• Journal of Conservation Science
• /
• v.31 no.3
• /
• pp.215-225
• /
• 2015

"Jocheonillgi", one of the articles left by Jo Heon(1544~1592) whose pen name was Jung-bong, is a travelogue which was written in around 1574~1575 when he was dispatched to Ming as a formal envoy(Giljungkwan). The diary-style manuscript was designated as one of the pieces of the National Treasure 1007. Due to the damage of its binding strings, abrasion, fold, stain, insect and damage on the surface, conservation and restoration was needed. In the process of separation, three more binding strings were found, confirming that the travelogue was rebound at least twice in the past. In addition, the page of 'Yeondoillgi', the original title, was bound inside, confirming that the current cover was revised in the past. As the result of the investigation of base fabric, paper mulberry was found to be the cover, inside paper, lining paper, paper strings. The three kinds of binding string including the current ones was defined to be silk.

• Journal of KIISE:Computer Systems and Theory
• /
• v.29 no.1
• /
• pp.16-21
• /
• 2002

Repetitive strings have been studied in such diverse fields as molecular biology data compression etc. Some important regularities that have been studied are perods, covers seeds and squares. A natural extension of the repetition problems is to allow errors. Among the four notions above aproximate squares and approximate periodes have been studied. In this paper, we introduce the notion of approximate covers which is an approximate version of covers. Given two strings P(｜P｜=m) and T(｜T｜=n) we propose and algorithm with finds the minimum distance t such that P is a t-approximate cover of T. The algorithm take O(m,n) time for the edit distance and $O(mn^2)$ time of finding a string which is an approximate cover of T is minimum distance is NP-complete.

• Journal of the Korean Society of Costume
• /
• v.66 no.5
• /
• pp.133-148
• /
• 2016

This study examined dahoe(多繪-braided cord) and mangsu(網綬-ornament of husu for ceremonial dress) used in myeon gwan(冕冠), daedae(大帶), and husu(後綬) among royal formal dresses in the Joseon Dynasty(1392-1910) based on historical materials including literature, relics, and paintings. The results of this study are as follows. In myeon-gwan, dahoe was used for cap strings, goeing(紘) and yeong(纓). Cap strings were applied to the king, the Crown Prince, and the eldest son of the Crown Prince regardless of their status, and they showed differences among the periods. Both goeing and yeong were used during the early period of Joseon, and then only yeong was used in the late period. As goeing was removed and only yeong was used in the late period, patterns combining goeing and yeong, in color and wearing method, appeared. Dahoe used in cap strings is dongdahoe(童多繪-a kind of braided cord). In daedae, 'nyuyak(紐約)' was tied up to its fastening part. The material of nyuyak was changed from dongdahoe in the early Joseon Dynasty to guangdahoe(廣多繪-a kind of braided cord) in the late period, and the method of using it was also changed. Husu was imported from Beijing in China during the early period of the Joseon Dynasty, but in 1747, it was regulated to be woven in Joseon, and at that time, King Yeongjo attempted to restore the institution of weaving husu with "320 su(首)," namely, 6,400 strands as specified for the status of a prince of the Ming Dynasty.

• Journal of KIISE:Software and Applications
• /
• v.30 no.1_2
• /
• pp.93-102
• /
• 2003

A slant correction method for handwritten Korean strings based on analysis of stroke distribution, which effectively reflects structural properties of Korean characters, is presented in this paper. The method aims to deal with typical problems which have been frequently observed in slant correction of handwritten Korean strings with conventional approaches developed for English/European languages. Extracted strokes from a line of text image are classified into two clusters by applying the K-means clustering. Gaussian modeling is applied to each of the clusters and the slant angle is estimated from the model which represents the vertical strokes. Experimental results support the effectiveness of the proposed method. For the performance comparison 1,300 handwritten address string images were used, and the results show that the proposed method has more superior performance than other conventional approaches.

• Journal of KIISE
• /
• v.42 no.10
• /
• pp.1247-1253
• /
• 2015

An inverted index structure is widely used for efficient string similarity search. One of the main requirements of similarity search is a fast response time; to this end, most techniques use an in-memory index structure. Since the size of an inverted index structure usually very large, however, it is not practical to assume that an index structure will fit into the main memory. To alleviate this problem, we propose a novel technique that reduces the size of an inverted index. In order to reduce the size of an index, the proposed technique rearranges data strings so that the data strings containing the same q-grams can be placed close to one other. Then, the technique encodes those multiple strings into a range. Through an experimental study using real data sets, we show that our technique significantly reduces the size of an inverted index without sacrificing query processing time.