• Proceedings of the IEEK Conference
• /
• 1999.06a
• /
• pp.496-499
• /
• 1999

This paper describes a medular neural network(MNN) for a vision system which tracks a given object using a sequence of images from a camera unit. The MNN is used to precisely recognize the given robot endeffector and to minize the processing time. Since the robot endeffector can be viewed in many different shapes in 3-D space, a MNN structure, which contains a set of feedforwared neural networks, co be more attractive in recognizing the given object. Each single neural network learns the endeffector with a cluster of training patterns. The training patterns for a neural network share the similar charateristics so that they can be easily trained. The trained MNN is less sensitive to noise and it shows the better performance in recognizing the endeffector. The recognition rate of MNN is enhanced by 14% over the single neural network. A vision system with the MNN can precisely recognize the endeffector and place it at the center of a display for a remote operator.

• International Journal of Advanced Culture Technology
• /
• v.6 no.3
• /
• pp.173-177
• /
• 2018

Masks are generated by adding different fonts of learning data characters in pixel unit, and pixel values belonging to each of the masks are divided into 3 groups. Using the directional expansion operators, we expand the text area of the test data character into 4 diagonal directions in order to create the boundary areas to distinguish it from the background area. A mask with a minimum average discordance is selected as the final recognition result by calculating the degree of discordance between the expanded test data and the masks. Image comparison using directional expansion operations more accurately recognizes test data through 4 subdivided recognition processes. It is also possible to expand the ranges of 3 groups of pixel values of masks more evenly such that new fonts can easily be added to the given learning data.

• Proceedings of the IEEK Conference
• /
• 2003.07e
• /
• pp.2419-2422
• /
• 2003

To improve performance of sentence speech recognition systems, we need to consider perplexity of language model and the number of words of dictionary for increasing vocabulary size. In this paper, we propose a language model of VCCV units for sentence speech recognition. For this, we choose VCCV units as a processing units of language model and compare it with clauses and morphemes. Clauses and morphemes have many vocabulary and high perplexity. But VCCV units have small lexicon size and limited vocabulary. An advantage of VCCV units is low perplexity. This paper made language model using bigram about given text. We calculated perplexity of each language processing unit. The perplexity of VCCV units is lower than morpheme and clause.

• Journal of the Korean Society of Industry Convergence
• /
• v.18 no.4
• /
• pp.241-249
• /
• 2015

The purpose of this research is to develop the pattern recognition algorithm based on smart camera for inspection automation, and including external surface state of molding parts or optical parts. By performance verification, this development can be applied to establish for existing reflex data because inputting surface badness degree of scratch's standard specification condition directly. And it is pdssible to distinguish from schedule error of badness product and normalcy product within schedule extent after calculating the error comparing actuality measurement reflex data and standard reflex data mutually. The proposed technology cab be applied to test for masearing of the smallest 10 pixel unit. It is illustrated the relibility pf proposed technology by an experiment.

• Proceedings of the Acoustical Society of Korea Conference
• /
• spring
• /
• pp.33-36
• /
• 2002

We implemented a robust Korean broadcast news transcription system for out-of-vocabulary (OOV), tested its performance. The occurrence of OOV words in the input speech is inevitable in large vocabulary continuous speech recognition (LVCSR). The known vocabulary will never be complete due to the existence of for instance neologisms, proper names, and compounds in some languages. The fixed vocabulary and language model of LVCSR system directly face with these OOV words. Therefore our Broadcast news recognition system has an offline OOV update module of language model and vocabulary to solve OOV problem and selects morpheme-based recognition unit (so called, pseudo-morpheme) for OOV robustness.

• The Journal of the Korea Contents Association
• /
• v.9 no.8
• /
• pp.57-63
• /
• 2009

Object recognition technologies using PCA(principal component analysis) recognize objects by deciding representative features of objects in the model image, extracting feature vectors from objects in a image and measuring the distance between them and object representation. Given frequent recognition problems associated with the use of point-to-point distance approach, this study adopted the k-nearest neighbor technique(class-to-class) in which a group of object models of the same class is used as recognition unit for the images in-putted on a continual input image. However, the robustness of recognition strategies using PCA depends on several factors, including illumination. When scene constancy is not secured due to varying illumination conditions, the learning performance the feature detector can be compromised, undermining the recognition quality. This paper proposes a new PCA recognition in which database of objects can be detected under different illuminations between input images and the model images.

• International Journal of Advanced Culture Technology
• /
• v.5 no.3
• /
• pp.67-72
• /
• 2017

Transform variation technique is constituted the vibration status of the flash-gap recognition level (FGRL) on the distribution recognition function. The recognition level condition by the distribution recognition function system is associated with the scattering vibration system. As to search a position of the dot model, we are consisted of the distribution value with character point by the output signal. The concept of recognition level is composed the reference of flash-gap level for variation signal by the distribution vibration function. For displaying a variation of the FGRL of the maximum-average in terms of the vibration function, and distribution position vibration that was the a distribution value of the far variation of the $Dis-rf-FA-{\alpha}_{MAX-AVG}$ with $5.74{\pm}1.12$ units, that was the a distribution value of the convenient variation of the $Dis-rf-CO-{\alpha}_{MAX-AVG}$ with $1.64{\pm}0.16$ units, that was the a distribution value of the flank variation of the $Dis-rf-FL-{\alpha}_{MAX-AVG}$ with $0.74{\pm}0.24$ units, that was the a distribution value of the vicinage variation of the $Dis-rf-VI-{\alpha}_{MAX-AVG}$ with $0.12{\pm}0.01$ units. The scattering vibration will be to evaluate at the ability of the vibration function with character point by the distribution recognition level on the FGRL that is showed the flash-gap function by the recognition level system. Scattering recognition system will be possible to control of a function by the special signal and to use a distribution data of scattering vibration level.

• Journal of Institute of Control, Robotics and Systems
• /
• v.21 no.1
• /
• pp.65-69
• /
• 2015

A vision-based blank alignment unit for a press automation line is introduced in this paper. A press is a machine tool that changes the shape of a blank by applying pressure and is widely used in industries requiring mass production. In traditional press automation lines, a mechanical centering unit, which consists of guides and ball bearings, is employed to align a blank before a robot inserts it into the press. However it can only align limited sized and shaped of blanks. Moreover it cannot be applied to a process where more than two blanks are simultaneously inserted. To overcome these problems, we developed a press centering unit by means of vision sensors for press automation lines. The specification of the vision system is determined by considering information of the blank and the required accuracy. A vision application S/W with pattern recognition, camera calibration and monitoring functions is designed to successfully detect multiple blanks. Through real experiments with an industrial robot, we validated that the proposed system was able to align various sizes and shapes of blanks, and successfully detect more than two blanks which were simultaneously inserted.

• Journal of the Korea Computer Industry Society
• /
• v.5 no.1
• /
• pp.65-76
• /
• 2004

In this paper, we observe the differences between linguistic and computational aspect in the automatic processing of languages which are dominant representation method for information in the Internet. For efficient information retrieval, information extraction and machine translation from the massive documents, we investigate analysis units for morphology analysis, syntactic analysis and semantic analysis. and propose the syntactic longest analysis unit rather than morphological unit based on linguistics. Also, by evaluating with massive documents, we show that the proposed analysis units can be used for the constraint which can reduce the ambiguity occurring in the language processing.

• Proceedings of the KSPS conference
• /
• 2006.05a
• /
• pp.117-120
• /
• 2006

This paper examined two hypotheses. Firstly, if the first syllable of word play an important role in visual word recognition, it may be the unit of word neighbor. Secondly, if the first syllable is the unit of lexical access, the neighborhood size effect and the neighborhood frequency effect would appear in a lexical decision task and a form primed lexical decision task. We conducted two experiments. Experiment 1 showed that words had large neighbors made a inhibitory effect in the LDT(lexical decision task). Experiment 2 showed the interaction between the neighborhood frequency effectand the word form similarity in the form primed LDT. We concluded that the first syllable in Korean words might be the unit of word neighborhood and play a central role in a lexical access.