• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.38 no.3
• /
• pp.269-281
• /
• 2001

This thesis deals with a novel font-dependent Hangul recognition algorithm invariant to position translation, scaling, and rotation using circular pattern vectors. The proposed algorithm removes noise from input letters using binary morphology and generates the circular pattern vectors. The generated circular pattern vectors represent spatial distributions on several concentric circles from the center of gravity in a given letter. Then the algorithm selects the letter minimizing the distance between the reference vectors and the generated circular pattern vectors. In order to estimate performances of the proposed algorithm, the completed Batang Hangul 2,350 letters were used as test images with scaling and rotational transformations. Experimental results show that the proposed algorithm are better than conventional algorithm using the ring projection in the recognition rates of Hangul letters with scaling and rotational transformation.

• ETRI Journal
• /
• v.26 no.6
• /
• pp.520-534
• /
• 2004

As the density of memories increases, unwanted interference between cells and the coupling noise between bit-lines become significant, requiring parallel testing. Testing high-density memories for a high degree of fault coverage requires either a relatively large number of test vectors or a significant amount of additional test circuitry. This paper proposes a new tiling method and an efficient built-in self-test (BIST) algorithm for neighborhood pattern-sensitive faults (NPSFs) and new neighborhood bit-line sensitive faults (NBLSFs). Instead of the conventional five-cell and nine-cell physical neighborhood layouts to test memory cells, a four-cell layout is utilized. This four-cell layout needs smaller test vectors, provides easier hardware implementation, and is more appropriate for both NPSFs and NBLSFs detection. A CMOS column decoder and the parallel comparator proposed by P. Mazumder are modified to implement the test procedure. Consequently, these reduce the number of transistors used for a BIST circuit. Also, we present algorithm properties such as the capability to detect stuck-at faults, transition faults, conventional pattern-sensitive faults, and neighborhood bit-line sensitive faults.

• Korean Journal of Microbiology
• /
• v.4 no.2
• /
• pp.19-21
• /
• 1966

For the purpose of determining possibility of aphid-transmission of mosaic disease of black locust, cowpea aphid (Aphis medicagnis Koch) and green peach aphid (Myzus persicae Sulzer) were experimented using cowpea as test plant, and both proved to be the vectors. As for transmission threshold period of cowpea aphid to the virus, the acquisition feeding period was five seconds and inoculation feeding period was two minutes. This black locust mosaic virus, therefore, is a nonpersistent virus.

• The Journal of Korean Society of Virology
• /
• v.29 no.1
• /
• pp.33-38
• /
• 1999

Retroviral vector provide a highly efficient method for gene transfer into eukaryotic cells. This vector system can be divided into two components; the retroviral vector itself and the retroviral packaging cell line. The key improvement in the design of these two components are, focused on two aspects; the reduction of helper virus production and high titer-virus. We used PA317 for retrovirus packaging cell line, for its high producibility of viral titer. To test the ability of the vectors to generate high titer-virus, we have chosen four different retroviral vectors; LN, LNSX, LNCX and LXSN. To test easily the viral titer, we have made recombinant construction with CD4 and CD8, checked their viral titer and stained their surface expression. LXSN which contain SV40 early promoter in front of neo gene showed best results in viral transient transfection assay, dot blot assay and surface expression. In addition, recombinant containing CD8 generally showed much higher viral titration and surface expression than CD4.

• Journal of Mechanical Science and Technology
• /
• v.17 no.3
• /
• pp.409-421
• /
• 2003

The hierarchical recursive local-correlation PIV algorithm with CBC (correlation based correction) method was employed to increase the spatial resolution of PIV results and to reduce error vectors. The performance of this new PIV algorithm was tested using synthetic images, PIV standard images of Visualization Society of Japan, real flows including ventilation flow inside a vehicle passenger compartment and wake behind a circular cylinder with riblet surface. As a result, most spurious vectors were suppressed by employing the CBC method, the hierarchical recursive correlation algorithm improved the sub-pixel accuracy of PIV results by decreasing the interrogation window size and Increased spatial resolution significantly. However, with recursively decreasing of interrogation window size, the SNR (signal-to-noise ratio) in the correlation plane was decreased and number of spurious vectors was increased. Therefore, compromised determination of optimal interrogation window size is required for given flow images, the performance of recursive algorithm is also discussed from a viewpoint of recovery ratio and error ratio in the paper.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 1998.04a
• /
• pp.244-249
• /
• 1998

The present paper improves the direct identification scheme based upon the equation error formulation with incomplete modal data. First, an indirect estimation technique is considered for estimating unmeasured elements of latent vectors by the combined use of a model and measured incomplete eigen vectors. It is used for estimating the other elements of eigen vectors, which are essential for identification but not available. Next an index is introduced here to indicate the quality of estimation with respect to the mode and the measured positions. A sensitivity formula for eigenvalues with respect to the unknown joint coefficient is also derived to select the modes appropriate for identification. An identification strategy is suggested to meet with practical problems with the help of the index and sensitivity formula. The index and the sensitivity are proved to be useful for selecting measurement positions and modes appropriate for identification A comprehensive simulation study is performed to test the proposed method.

• Proceedings of the IEEK Conference
• /
• 2000.07b
• /
• pp.843-846
• /
• 2000

An optical flow-based motion estimation algorithm is proposed for video coding. The algorithm uses block-matching motion estimation with an adaptive search region. The search region is computed from motion fields that are estimated based on the optical flow. The algorithm is based on the fact that true block-motion vectors have similar characteristics to optical flow vectors. Thereafter, the search region is computed using these optical flow vectors that include spatial relationships. In conventional block matching, the search region is fixed. In contrast, in the new method, the appropriate size and location of the search region are both decided by the proposed algorithm. The results obtained using test images show that the proposed algorithm can produce a significant improvement compared with previous block-matching algorithms.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.1C
• /
• pp.37-47
• /
• 2009

When H.264 encoded video streams are transmitted over wireless network, packet loss is unavoidable. Responding on this environment, we propose methods to recover missed motion vector in the decoder: At first, A candidate vector set for missing macroblock is estimated from high correlation coefficient of neighboring motion vectors and missing block vectors the algorithm clusters candidate vectors through distances amongst motion vectors of neighboring blocks. Then the optimal candidate vector is determined by the median value of the clustered motion vector set. In next stage, from the candidate vector set, the final candidate vector of missing block is determined it has minimum distortion value considering directions of neighboring pixels' boundary. Test results showed that the proposed algorithm decreases the candidate motion vectors $23{\sim}61%$ and reduces $3{\sim}4sec$ on average processing(decoding) time comparing the existing H.264 codec. The PSNR, in terms of visual quality is similar to existing methods.

• Journal of KIISE:Software and Applications
• /
• v.33 no.10
• /
• pp.836-850
• /
• 2006

Nowadays, due to the rapidly increasing number of package software products, quality test has been emphasized for package software products. When testing software products, one of the most important factors is to select metrics which form the bases for tests. In this paper, the types of package software are represented as characteristic vectors having probabilistic relationships with metrics. The characteristic vectors could be regarded as indicators of software type. To assign the metrics for each software type, the past test metrics are collected and analyzed. Using Bayesian belief network, the dependency relationship network of the characteristic vectors and metrics is constructed. The dependency relationship network is then used to find the proper metrics for the test of new package software products.

• Journal of Korean Institute of Industrial Engineers
• /
• v.40 no.4
• /
• pp.404-414
• /
• 2014

Yield prediction is important to manage semiconductor quality. Many researches with machine learning algorithms such as SVM (support vector machine) are conducted to predict yield precisely. However, yield prediction using SVM is hard because extremely imbalanced and big data are generated by final test procedure in semiconductor manufacturing process. Using SVM algorithm with imbalanced data sometimes cause unnecessary support vectors from major class because of unselected support vectors from minor class. So, decision boundary at target class can be overwhelmed by effect of observations in major class. For this reason, we propose a under-sampling method with minor class based SVM (MCSVM) which overcomes the limitations of ordinary SVM algorithm. MCSVM constructs the model that fixes some of data from minor class as support vectors, and they can be good samples representing the nature of target class. Several experimental studies with using the data sets from UCI and real manufacturing process represent that our proposed method performs better than existing sampling methods.