• Smart Media Journal
• /
• v.5 no.2
• /
• pp.59-64
• /
• 2016

Storytelling techniques have been using for utilizing in various fields such as education, marketing and so on. In creative content sector, especially, the amount of creating the a newly story content from existing oral tales which are short and quite hard to understand the context has been increased. However, there are some limitations for creating the ideas of the story according to the expansion of the stories while the approach for new adaptation of existing stories is quite challenging. In this paper, therefore, we described the preliminary steps of text analysis and object extraction method for story development which can be applied to the storytelling authoring supported software.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.7
• /
• pp.528-538
• /
• 2003

A SCCRS (Speech and Character Combined Recognition System) is developed for working on mobile devices such as PDA (Personal Digital Assistants). In SCCRS, the feature extraction is separately carried out for speech and for hand-written character, but the recognition is performed in a common engine. The recognition engine employs essentially CHMM (Continuous Hidden Markov Model), which consists of variable parameter topology in order to minimize the number of model parameters and to reduce recognition time. For generating contort independent variable parameter model, we propose the SSMS(Successive State and Mixture Splitting), which gives appropriate numbers of mixture and of states through splitting in mixture domain and in time domain. The recognition results show that the proposed SSMS method can reduce the total number of GOPDD (Gaussian Output Probability Density Distribution) up to 40.0% compared to the conventional method with fixed parameter model, at the same recognition performance in speech recognition system.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.36C no.1
• /
• pp.73-81
• /
• 1999

This paper describes the methods which segment more efficiently the car license plate and the character region by using 1-D DCT. In the car images, a license plate region and a character region of the license plate can be distingushed by the regular high frequency components from the car images. In this method, it is shown that the regular high frequency componets are extracted by using DCT and license plate region is segmented in the car image and the caracter region is then seperated at the extracted license plate by using the previously extracted regular high frequency components. Some experiment results of the various images are shown. It has been shown from the results that the car license plates and the character regions can be segmented more exactly and efficiently than conventional methods.

• Journal of Internet Computing and Services
• /
• v.6 no.4
• /
• pp.93-102
• /
• 2005

In Recent times, Netizens have frequently use virtual character 'Avatar' schemes in order to present their own identity, there is a strong need for avatars to resemble the user. This paper proposes an extraction technique for facial region and features that are used in generating the avatar automatically. For extraction of facial feature component, the method uses ACM and edge information. Also, in the extraction process of facial region, the proposed method reduces the effect of lights and poor image quality on low resolution pictures. this is achieved by using the variation of facial area size which is employed for external energy of ACM. Our experiments show that the success rate of extracting facial regions is $92{\%}$ and accuracy rate of extracting facial feature components is $83.4{\%}$, our results provide good evidence that the suggested method can extract the facial regions and features accurately, moreover this technique can be used in the process of handling features according to the pattern parts of automatic avatar generation system in the near future.

• Journal of the Korea Convergence Society
• /
• v.9 no.11
• /
• pp.9-14
• /
• 2018

The vision system is a device for acquiring images and analyzing and discriminating inspection areas. Demand for use in the automation process has increased, and the introduction of a vision-based inspection system has emerged as a very important issue. These vision systems are used for everyday life and used as inspection equipment in production processes. Image processing technology is actively being studied. However, there is little research on the area definition for extracting objects such as character recognition or semiconductor packages. In this paper, define a region of interest and perform edge extraction to prevent the user from judging noise as an edge. We propose a noise-robust alignment correction model that can extract the edge of a region to be inspected using the distribution of edges in a specific region even if noise exists in the image. Through the proposed model, it is expected that the product production efficiency will be improved if it is applied to production field such as character recognition of tire or inspection of semiconductor packages.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.380-382
• /
• 2021

According to the National Police Agency, 161 out of 38,496 unsolved cases as of 2020. Most of the adult missing persons, the highest of the unsolved causes, are evaluated as simple runaway, which takes a long time to investigate. Even if search through CCTV, it can take a long time and the accuracy can be somewhat low because you have to check the faces of the characters one by one and find the characters only with the characteristics of the statements. This paper utilizes MTCNN to conduct research on character extraction in CCTV. We initiate simultaneous analysis of the features of faces learned with MTCNN and the clothes we are wearing, so that only the overlapping characters are extracted so that they can be identified to the related parties. For aim to learn more diverse feature detection to narrow down the features of missing persons in the future and increase their accuracy.

• 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 Korea Multimedia Society
• /
• v.15 no.6
• /
• pp.737-747
• /
• 2012

Recently, in order to support automatic meter reading for conventional metering devices, an image processing-based approach of recognizing the number meter data in the captured meter images has attracted many researchers' interests. Numerical character segmentation is a very critical process for successful recognition. In this paper, we propose an efficient numeric character segmentation method which can segment numeric characters well for any metering device types under diverse illumination environments. The proposed method consists of two consecutive stages; detection of number area containing all numbers as a tight ROI(Region of Interest) and segmentation of numerical characters in the ROI. Detection of tight ROI is achieved in two steps: extraction of rough ROI by utilizing horizontal line segments after illumination enhancement preprocessing, and making the rough ROI more tight through clipping utilizing vertical and horizontal projection about binarized ROI. Numerical character segmentation in the detected ROI is stably achieved in two processes of 'vertical segmentation of each number region' and 'number segmentation in the each vertical segmented number region'. Through the experiments about a homegrown meter image database containing various meter type images of low contrast, low intensity, shadow, and saturation, it is shown that the proposed numeric character segmentation method performs effectively well for any metering device types under diverse illumination environments.

• The Journal of the Convergence on Culture Technology
• /
• v.4 no.3
• /
• pp.241-246
• /
• 2018

Because of the mechanical problems of input camera equipment, image restoration process is performed in order to minimize recognition errors due to the noise problem generated in test data image. The image restoration method resolves the noise problem by examining the numbers and positions of the Direct neighbors and the Indirect neighbors for each pixel constituting the test data. As a result, satisfactory recognition result can be obtained by eliminating the noise problem generated in the test data through the image restoration process as much as possible and also by calculating the differences between the learning data and the test data in the area unit, thereby reducing the possibility of recognition error by the noise problem.

• Journal of Environmental Health Sciences
• /
• v.27 no.1
• /
• pp.8-13
• /
• 2001

The tiles were manufactured using a mixture of the TK material(a raw material in making tile) and Cu-Cr-Ag impregnated activated carbon(ASC Charcoal). The extraction character of heavy metals in making tile was evaluated and the manufacturing conditions of tile were studied. The heavy metals in the mixture-before and after the tile was production of tiles was successful and as a result of heavy metal analysis, the tile showed that the concentration of heavy metal after the production of tiles was lower than that of the before one. The concentration of eluted heavy metal by acidic and basic solutions was low and the quality of the produced tile was similar to the commercial one. The result of this study suggested that the waste ASC charcoal was used to produce good tiles and it also might reduce soil pollution.