• Food Engineering Progress
• /
• v.15 no.1
• /
• pp.48-55
• /
• 2011

In this research, the near-infrared absorption from 1,100-2,300 nm was used to measure the content of capsaicinoids in the red-pepper powder by using the Acousto-optic tunable filters (AOTF) spectrometer with sample plate and sample rotating unit. Non-spicy red-pepper samples from one location (Younggwang-gun. Korea) were mixed with spicy one (var. Chungyang) to make samples separated by particle size (below 0.425 mm, 0.425-0.71 mm, and 0.71- 1.4 mm). The Partial Least Squares Regression (PLSR) model to predict the capsaicinoid content on particle sizes was developed with measured spectra by AOTF spectrometer and used to analyze the amount of capsaicinoids by HPLC. The PLSR Model of red-pepper powder of below 0.425 mm, 0.425-0.71 mm, and 0.71-1.4 mm with cross validation had ${R_V}^2$ = 0.948-0.979 and Standard Error of Prediction (SEP) = 6.56-7.94 mg%. The prediction error of smaller particle size of red-pepper powder was low. The best PLSR model was found in pretreatment of Range Normalization, Standard Normal Variate, and 1st Derivatives of red-pepper powder of below 1.4 mm with cross validation, having ${R_V}^2$ = 0.959 and SEP = 8.82 mg%.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.

• Journal of KIISE:Software and Applications
• /
• v.30 no.9
• /
• pp.829-842
• /
• 2003

In this Paper, we propose the new face detection and tracking method based on template matching for real-time applications such as, teleconference, telecommunication, front stage of surveillance system using face recognition, and video-phone applications. Since the main purpose of paper is to track a face regardless of various environments, we use template-based face tracking method. To generate robust face templates, we apply wavelet transform to the average face image and extract three types of wavelet template from transformed low-resolution average face. However template matching is generally sensitive to the change of illumination conditions, we apply Min-max normalization with histogram equalization according to the variation of intensity. Tracking method is also applied to reduce the computation time and predict precise face candidate region. Finally, facial components are also detected and from the relative distance of two eyes, we estimate the size of facial ellipse.

• Journal of the Korean earth science society
• /
• v.28 no.3
• /
• pp.311-323
• /
• 2007

The Yellow and East China seas received a vast amount of sediment $(>10^9ton/yr)$, which comes mainly from the Changjiang and Huanghe rivers of China and the Korean rivers. However, there are still no direct sedimentological-geochemical indicators, which can distinguish these two end-members (Korean and Chinese river sources) in these seas. The purpose of this study is to provide the potential geochemical-tracers enabling these river materials to be identified within the sediment load of the Yellow and East China seas. The compositions of major elements (Al, Fe, Mg, K, Ca, Na, and Ti) of Chinese and Korean river sediments were analyzed. To minimize the grain-size effect, furthermore, bulk sediments were separated into two groups, silt $(60-20{\mu}m)$ and clay $(<20{\mu}m)$ fractions, and samples of each fraction were analyzed for major and strontium isotope $(^{87}Sr/^{86}Sr)$ compositions. In this study, Fe/Al and Mg/Al ratios in bulk sediment samples, using a new Al-normalization procedure, are suggested as an excellent tool for distinguishing the source of sediments in the Yellow and East China seas. This result is clearly supported by the concentrations of these elements in silt and clay fraction samples. In silt fraction samples, Korean river sediments have much higher $^{87}Sr/^{86}Sr$ ratio $(0.7229{\sim}0.7253)$ than Chinese river sediments $(0.7169{\sim}0.7189)$, which suggests the distribution pattern of $^{87}Sr/^{86}Sr$ ratios as a new tracer to discriminate the provenance of shelf sediments in the Yellow and East China seas. On the basis of these geochemical tracers, clay fractions of southeastern Yellow Sea mud (SEYSM) patch may be a mixture of two sediments originated from Korea and China. In contrast, the geochemical compositions of silt fractions are very close to that of Korea river sediments, which indicates that the silty sediments of SEYSM are mainly originated from Korean rivers.

• Korean Journal of Cognitive Science
• /
• v.13 no.3
• /
• pp.1-10
• /
• 2002

This study was performed to analyze the physical attributes of the faces and affective words on the fares. For analyzing physical attributes inside of a face, 36 facial features were selected and almost of them were the lengths or distance values. For analyzing facial contour 14 points were selected and the lengths from nose-end to them were measured. The values of these features except ratio values normalized by facial vortical length or facial horizontal length because the face size of each person is different. The principal component analysis (PCA) was performed and four major factors were extracted: 'facial contour' component, 'vortical length of eye' component, 'facial width' component, 'eyebrow region' component. We supposed the five-dimensional imaginary space of faces using factor scores of PCA, and selected representative faces evenly in this space. On the other hand, the affective words on faces were collected from magazines and through surveys. The factor analysis and multidimensional scaling method were performed and two orthogonal dimensions for the affections on faces were suggested: babyish-mature and sharp-soft.

• Journal of Korean Society of Environmental Engineers
• /
• v.31 no.2
• /
• pp.153-160
• /
• 2009

A PM10 (aerodynamic diameter${\leq}$10 ${\mu}m$) sampler is used to quantify the potential human exposure to suspended particulate matter (PM) and to comply with the governmental regulation. This study was conducted to compare and evaluate the same PM10 cutpoint and different slopes between United States Environmental Protection Agency (USEPA) PM10 sampling criterion and American Conference of Governmental Industrial Hygienists/$Comit\acute{e}$ $Europ\acute{e}en$ de Normalization/International Organization for Standardization thoracic PM10 sampling criterion through theory and experiment. Four PM10 samplers according to the USEPA criterion and one RespiCon sampler in accordance with the thoracic PM10 criterion were used in the present study. In addition, one DustTrak monitor was used to measure real time PM10 mass concentrations. All six aerosol samplers were tested in a PM generation chamber using polydisperse fly ash. Theoretical mass concentrations were calculated by applying the measured particle size distribution characteristics (geometric mean = 6.6 ${\mu}m$, geometric standard deviation = 1.9) of fly ash to each sampling criterion. The measured mass concentrations through a chamber experiment were consistent with theoretical mass concentrations in that a RespiCon sampler with the thoracic PM10 criterion collected less PM than a PM10 sampler with the USEPA criterion. The overall chamber experiment results indicated, when a PM10 sampler was used as a reference sampler, that (1) a RespiCon sampler had a normalizing factor of 1.6, meaning that this sampler underestimated an average 60% of PM10 mass sampled from a PM10 sampler, and (2) a DustTrak real-time monitor using a PM10 inlet had a calibration factor of 2.1.

• Journal of Radiation Protection and Research
• /
• v.30 no.1
• /
• pp.1-7
• /
• 2005

Radiation absorption parameters of carbon fiber panel were measured in comparison to acrylic panel. $30{\times}30cm$ sized 2mm thick carbon fiber panel and identical sized 6mm thick acrylic panel were placed in tray holder position and 0cm, 5cm, 10cm from surface of phantom. Radiation field size was $10{\times}10cm$. 50MU of 4MV photon was irradiated to the phantom with dose rate of 300MU/min. Source-to-phantom distance was 120cm. Radiation dose was measured with 0.6cc Farmer-type ionization chamber with 1cm build-up. Measurement was repeated thrice and normalization was done to the dose of the open field. Radiation transmission rate of carbon fiber panel is approximately 1% lower than acrylic panel of equivalent thickness. However, considering the strength of the material, transmission rate is higher for carbon fiber panel. Although carbon fiber panel increases the radiation dose when attached to the surface for about 2%, it normalizes the radiation dose to 97-99% of irradiated dose which could have been lowered to as much as 5-7.5% with acrylic panel. As carbon fiber panel is stronger than acrylic panel, radiation fixation device could be made thinner and thus lighter and furthermore, with increased radiation transmission. This in turn makes carbon fiber more ideal material for radiation fixation device over conventionally used acrylic.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.4
• /
• pp.189-198
• /
• 2024

Lipreading is one of the important parts of speech recognition, and several studies have been conducted to improve the performance of lipreading in lipreading systems for speech recognition. Recent studies have used method to modify the model architecture of lipreading system to improve recognition performance. Unlike previous research that improve recognition performance by modifying model architecture, we aim to improve recognition performance without any change in model architecture. In order to improve the recognition performance without modifying the model architecture, we refer to the cues used in human lipreading and set other regions such as chin and cheeks as regions of interest along with the lip region, which is the existing region of interest of lipreading systems, and compare the recognition rate of each region of interest to propose the highest performing region of interest In addition, assuming that the difference in normalization results caused by the difference in interpolation method during the process of normalizing the size of the region of interest affects the recognition performance, we interpolate the same region of interest using nearest neighbor interpolation, bilinear interpolation, and bicubic interpolation, and compare the recognition rate of each interpolation method to propose the best performing interpolation method. Each region of interest was detected by training an object detection neural network, and dynamic time warping templates were generated by normalizing each region of interest, extracting and combining features, and mapping the dimensionality reduction of the combined features into a low-dimensional space. The recognition rate was evaluated by comparing the distance between the generated dynamic time warping templates and the data mapped to the low-dimensional space. In the comparison of regions of interest, the result of the region of interest containing only the lip region showed an average recognition rate of 97.36%, which is 3.44% higher than the average recognition rate of 93.92% in the previous study, and in the comparison of interpolation methods, the bilinear interpolation method performed 97.36%, which is 14.65% higher than the nearest neighbor interpolation method and 5.55% higher than the bicubic interpolation method. The code used in this study can be found a https://github.com/haraisi2/Lipreading-Systems.