• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.1031-1033
• /
• 2005

This paper presents a system of handwritten numerals recognition, which is based on Ant-miner algorithm (data mining based on Ant colony optimization). At the beginning, three distinct fractures (also called attributes) of each numeral are extracted. The attributes are Loop zones, End points, and Feature codes. After these data are extracted, the attributes are in the form of attribute = value (eg. End point10 = true). The extraction is started by dividing the numeral into 12 zones. The numbers 1-12 are referenced for each zone. The possible values of Loop zone attribute in each zone are "true" and "false". The meaning of "true" is that the zone contains the loop of the numeral. The Endpoint attribute being "true" means that this zone contains the end point of the numeral. There are 24 attributes now. The Feature code attribute tells us how many lines of a numeral are passed by the referenced line. There are 7 referenced lines used in this experiment. The total attributes are 31. All attributes are used for construction of the classification rules by the Ant-miner algorithm in order to classify 10 numerals. The Ant-miner algorithm is adapted with a little change in this experiment for a better recognition rate. The results showed the system can recognize all of the training set (a thousand items of data from 50 people). When the unseen data is tested from 10 people, the recognition rate is 98 %.

• Journal of IKEEE
• /
• v.22 no.3
• /
• pp.578-584
• /
• 2018

Image segmentation has been presented in the various method in image interpretation and recognition, and the image is using separate the characteristics of the specific purpose. In this paper, we proposed an algorithm that separate image for feature points detected to high temperature in a Thermal infrared image. In order to improve the processing time, the proposed algorithm is implemented to FPGA Hardware Block using the Zynq-7000 Evaluation Board environment. The proposed High-Temperature Detection Algorithm and total FPGA blocks show a decrease of a processing time result from 16ms to 0.001ms, and from 50ms to 0.322ms respectively. It is also verified similar results of the PSNR to comparing software thermal testbench and hardware ones.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.13 no.1
• /
• pp.58-63
• /
• 2020

This paper proposes the method for improving the localization accuracy of the mobile robot based on the stereo camera. To restore the position information from stereo images obtained by the stereo camera, the corresponding point which corresponds to one pixel on the left image should be found on the right image. For this, there is the general method to search for corresponding point by calculating the similarity of pixel with pixels on the epipolar line. However, there are some disadvantages because all pixels on the epipolar line should be calculated and the similarity is calculated by only pixel value like RGB color space. To make up for this weak point, this paper implements the method to search for the corresponding point simply by calculating the gap of x-coordinate when the feature points, which are extracted by feature extraction and matched by feature matching method, are a pair and located on the same y-coordinate on the left/right image. In addition, the proposed method tries to preserve the number of feature points as much as possible by finding the corresponding points through the conventional algorithm in case of unmatched features. Because the number of the feature points has effect on the accuracy of the localization. The position of the mobile robot is compensated based on 3-D coordinates of the features which are restored by the feature points and corresponding points. As experimental results, by the proposed method, the number of the feature points are increased for compensating the position and the position of the mobile robot can be compensated more than only feature extraction.

• Journal of Biomedical Engineering Research
• /
• v.16 no.2
• /
• pp.121-128
• /
• 1995

Information about retinal blood vessels can be used in grading disease severity or as part of the process of automated diagnosis of diseases with ocular menifestations. In this paper, we address the problem of detecting retinal blood vessels and optic disk (papilla) in eye-fundus images. We introduce an algorithm for feature extraction based on Fuzzy Clustering algorithm (fuzzy c-means). A method of finding the optic disk (papilla) is proposed in the eye-fundus images. Additionally, the inrormations such as position and area of the optic disk are extracted. The results are compared to those obtained from other methods. The automatic detection of retinal blood vessels and optic disk in the eye-rundus images could help physicians in diagnosing ocular diseases.

• Proceedings of the IEEK Conference
• /
• 2002.06c
• /
• pp.83-86
• /
• 2002

This paper intends to provide convenience to the disabled, losing the use of their limbs, through voice recognition technology. The voice recognition part of this system recognizes voice by DTW (Dynamic Time Warping) Which is most Widely used in Speaker dependent system. Specially, S/N rate was improved through Wiener filter in the pre-treatment phase while considering real environmental conditions; the result values of 12th order feature pattern per frame are extracted by DTW algorithm using LPC and Cepsturm in feature extraction process. Furthermore, miniaturization is pursued using TMS320C32, 71's the floating-point DSP, for the hardware part. Currently, 90% of hardware porting has been completed, but we can confirm that the recognition rate was 96% as a result of performing the DTW algorithm in PC.

• Proceedings of the KSME Conference
• /
• 2004.04a
• /
• pp.822-827
• /
• 2004

In this paper, an intelligent fault diagnosis system is proposed for induction motors through the combination of feature extraction, genetic algorithm (GA) and neural network (ANN) techniques. Features are extracted from motor vibration signals, while reducing data transfers and making on-line application available. GA is used to select most significant features from whole feature database and optimize the ANN structure parameter. Optimized ANN diagnoses the condition of induction motors online after trained by the selected features. The combination of advanced techniques reduces the learning time and increases the diagnosis accuracy. The efficiency of the proposed system is demonstrated through motor faults of electrical and mechanical origin on the induction motors. The results of the test indicate that the proposed system is promising for real time application.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.34C no.1
• /
• pp.51-58
• /
• 1997

In passive sonar, the frequency spectrum of a sound radiated by underwater moving targets is composed of a broadband nonuniform background noise and narrowband discrete tonals. To detect the tonals, the background noise is estimated and removed. Using the existing algorithms that estimate the background noise, a week tonals are not detected. Because a freuqency line that is formed by tonals which are being extracted continuously is a feture of the target, we are nessesory to efficiently detect the tonals that compose the frequncy line. In this paper, we propose an efficient neural network that can remove automatically the background and detect the even errl tonals, and we extract the frequency line feature on the spectrogram by the proposed algorithm. The experimental results for a ship's radiated sound show a better performance in comparison with the existing TPM algorithm.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.22 no.49
• /
• pp.11-21
• /
• 1999

The purpose of this study is to present the method to select optimal size for the industrial products which are closely related to human's body size. For this purpose, human factors such as body characteristics, body features, and preference in product selection which needs to be considered in setting standards were analyzed. This analysis is to select optimal size to minimize losses caused by the difference of size between demand by the customers and supply from the manufacturers. Using loss function, repetitive calculation process algorithm by using bisearch method was applied in selecting the sizes of demand and supply which minimize the total expected losses. For cumulative normal distribution probability, IMSL routine DNORDF was used. In case study, comparison has been made between the result which was calculated using presented algorithm and the results calculated by the process currently used by KS and ISO by measuring aged women's body size in human factors side and sorting them through the factor analysis and cluster analysis for feature factor extraction. Thus, they can be used as a basis for establishing industrial product standards.

• Proceedings of the IEEK Conference
• /
• 2009.05a
• /
• pp.222-224
• /
• 2009

This paper presents an algorithm that extracts road lane's specific information by using HSI color information and performance enhancement of lane detection base on vision processing of drive assist. As a preprocessing for high speed lane detection, the optimal extraction of region of interest for lane boundary(ROI-LB) can be processed to reduction of detection region in which high speed processing is enabled and it also increases reliabilities by deleting edges those are misrecognized. Road lane is extracted with simultaneous processing of noise reduction and edge enhancement using the Laplacian filter, the reliability of feature extraction can be increased for various road lane patterns. Since noise can be removed by using saturation and brightness of HSI color model. Also it searches for the road lane's color information and extracts characteristics. The real road experimental results are presented to evaluate the effectiveness of the proposed method.

• Journal of Information Processing Systems
• /
• v.20 no.3
• /
• pp.360-374
• /
• 2024

A personalized course recommendation algorithm based on deep learning in an online education cloud platform is proposed to address the challenges associated with effective information extraction and insufficient feature extraction. First, the user potential preferences are obtained through the course summary, course review information, user course history, and other data. Second, by embedding, the word vector is turned into a low-dimensional and dense real-valued vector, which is then fed into the compressed interaction network-deep neural network model. Finally, considering that learners and different interactive courses play different roles in the final recommendation and prediction results, an attention mechanism is introduced. The accuracy, recall rate, and F1 value of the proposed method are 0.851, 0.856, and 0.853, respectively, when the length of the recommendation list K is 35. Consequently, the proposed strategy outperforms the comparison model in terms of recommending customized course resources.