• Journal of Mechanical Science and Technology
• /
• v.19 no.10
• /
• pp.1856-1863
• /
• 2005

In this study, a visual sensor system for weld seam tracking the I-butt weld joints in GMA welding was constructed. The sensor system consists of a CCD camera, a diode laser with a cylindrical lens and a band-pass-filter to overcome the degrading of image due to spatters and arc light. In order to obtain the enhanced image, quantitative relationship between laser intensity and iris opening was investigated. Throughout the repeated experiments, the shutter speed was set at 1/1000 second for minimizing the effect of spatters on the image, and therefore the image without the spatter traces could be obtained. Region of interest was defined from the entire image and gray level of the searched laser stripe was compared to that of weld line. The differences between these gray levels lead to spot the position of weld joint using central difference method. The results showed that, as long as weld line is within $\pm15^{o}$ from the longitudinal straight line, the system constructed in this study could track the weld line successfully. Since the processing time is no longer than 0.05 sec, it is expected that the developed method could be adopted to high speed welding such as laser welding.

• Journal of Information Processing Systems
• /
• v.9 no.3
• /
• pp.349-364
• /
• 2013

The big challenge in current content-based image retrieval systems is to reduce the semantic gap between the low level-features and high-level concepts. In this paper, we have proposed a novel framework for efficient image retrieval to improve the retrieval results significantly as a means to addressing this problem. In our proposed method, we first extracted a strong set of image features by using the dual-tree rotated complex wavelet filters (DT-RCWF) and dual tree-complex wavelet transform (DT-CWT) jointly, which obtains features in 12 different directions. Second, we presented a relevance feedback (RF) framework for efficient image retrieval by employing a support vector machine (SVM), which learns the semantic relationship among images using the knowledge, based on the user interaction. Extensive experiments show that there is a significant improvement in retrieval performance with the proposed method using SVMRF compared with the retrieval performance without RF. The proposed method improves retrieval performance from 78.5% to 92.29% on the texture database in terms of retrieval accuracy and from 57.20% to 94.2% on the Corel image database, in terms of precision in a much lower number of iterations.

• Journal of Information Processing Systems
• /
• v.11 no.3
• /
• pp.421-437
• /
• 2015

This paper aims to present a supervised classification algorithm based on data fusion for the segmentation of the textured images. The feature extraction method we used is based on discrete wavelet transform (DWT). In the segmentation stage, the estimated feature vector of each pixel is sent to the support vector machine (SVM) classifier for initial labeling. To obtain a more accurate segmentation result, two strategies based on information fusion were used. We first integrated decision-level fusion strategies by combining decisions made by the SVM classifier within a sliding window. In the second strategy, the fuzzy set theory and rules based on probability theory were used to combine the scores obtained by SVM over a sliding window. Finally, the performance of the proposed segmentation algorithm was demonstrated on a variety of synthetic and real images and showed that the proposed data fusion method improved the classification accuracy compared to applying a SVM classifier. The results revealed that the overall accuracies of SVM classification of textured images is 88%, while our fusion methodology obtained an accuracy of up to 96%, depending on the size of the data base.

• Journal of the Korean Society for Precision Engineering
• /
• v.21 no.9
• /
• pp.32-40
• /
• 2004

This paper deals with an automated computer-aided process planning and die design system by which designer can determine operation sequences even if they have a little experience in process planning and die design for axisymmetric products. An attempt is made to link programs incorporating a number of expert design rules with the process variables obtained by commercial FEM softwares, DEFORM and ANSYS, to form a useful package. The system is composed of four main modules. The process planning and the die design modules consider several factors, such as the complexities of preform geometry, punch and die profiles, specifications of available multi former, and the availability of standard parts. They can provide a flexible process based on either the reduction in the number of forming sequences by combining the possible two processes in sequence, or the reduction of deviation of the distribution on the level of the required forming loads by controlling the forming ratios. Especially in die design module an optimal design technique and horizontal split die were investigated for determining appropriate dimensions of components of multi-former die set. It is constructed that the proposed method can be beneficial for improving the tool life of die set at practice.

• Journal of Korea Multimedia Society
• /
• v.13 no.5
• /
• pp.702-709
• /
• 2010

The performance of mass segmentation is greatly influenced by an initial position of a mass. Some researchers performed mass segmentation with the initial position of a mass given by radiologists. The purpose of our research is to find the initial position for mass segmentation and to notify the segmented mass to radiologists without any additional information on mammograms. The proposed system consists of breast segmentation by region growing and opening operations, decision of an initial seed with characteristics of masses, and mass segmentation by a level set segmentation. A seed for mass segmentation is set based on mass scoring measure calculated by block-based variances and masked information in a sub-sampled mammogram. We used a DDSM database to evaluate the system. The accuracy of mass detection is 78% sensitivity at 4 FP/image, and it reached 92% if multiple views for masses were considered.

• Proceedings of the Korean Society for Technology of Plasticity Conference
• /
• 2008.10a
• /
• pp.158-161
• /
• 2008

In the line array roll set (LARS) process, the initial plate is progressed into the final shape in a stepwise or pathwise manner according to the basic principle of the incremental forming process. The deformation proceeds simultaneously in the longitudinal and transverse directions. It was found that the curvature level of the formed plates in the previous study was well over the curvature required in shipyards. This fact shows that the LARS method has considerable potential for shipbuilding applications. In this study, several experiments with the LARS system is carried out fur manufacturing of plates with large curvatures. The bulbs at a stem and stern among ship hull plates correspond to these plates. Furthermore, the qualities of formed plates are evaluated according to the types of roll arrangements through experimental and numerical analyses.

• Journal of Biosystems Engineering
• /
• v.22 no.1
• /
• pp.30-40
• /
• 1997

A computer vision system was built to automatically grade the leaf tobacco. A color image processing algorithm was developed to extract shape, color and texture features. An improved back propagation algorithm in an artificial neural network was applied to grade the Burley type dried leaf tobacco. The success rate of grading in three-grade classification(1, 3, 5) was higher than the rate of grading in six-grade classification(1, 2, 3, 4, 5, off), on the average success rate of both the twenty-five local pixel-set and the sixteen local pixel-set. And, the average grading success rate using both shape and color features was higher than the rate using shape, color and texture features. Thus, the texture feature obtained by the spatial gray level dependence method was found not to be important in grading leaf tobacco. Grading according to the shape, color and texture features obtained by machine vision system seemed to be inadequate for replacing manual grading of Burely type dried leaf tobacco.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.40 no.6
• /
• pp.213-224
• /
• 2003

The proposed algorithm can always set up the optimal threshold value regardless of channel characteristics using an adaptive threshold establishment method that determines the threshold level according to channel noise power, and then it uses the specially designed training symbols that can make the algorithm's estimation performance be less sensitive to power delay profile variation in a multipath channel. In result, the estimation performance of the proposed technique is less affected by channel characteristic variation.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.11 no.4
• /
• pp.737-744
• /
• 2007

It need estimation model who is efficient and estimate correctly organization's information security level to achieve effectively organization's information security target. Also, estimate class information security level for this and need reformable estimation indicator or standard and estimation methodology of information security systems that application is possible should be studied in our country. Therefore many research centers including ISO are preparing the measuring and evaluating method for network duality. This study will represent an evaluating model for network security based on checklist. In addition, we propose ah measuring and evaluating method for network performance. The purpose of two studies is to present the evaluating procedure and method for measuring security of network on set workwill be identified and a measuring method and procedure will be proposed.

• The Journal of the Acoustical Society of Korea
• /
• v.18 no.3E
• /
• pp.3-8
• /
• 1999

In speech recognition for real-world applications, the performance degradation due to the mismatch introduced between training and testing environments should be overcome. In this paper, to reduce this mismatch, we provide a hybrid method of spectral subtraction and residual noise masking. We also employ multiple model approach to obtain improved robustness over various noise environments. In this approach, multiple model sets are made according to several noise masking levels and then a model set appropriate for the estimated noise level is selected automatically in recognition phase. According to speaker independent isolated word recognition experiments in car noise environments, the proposed method using model sets with only two masking levels reduced average word error rate by 60% in comparison with spectral subtraction method.