• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.8
• /
• pp.232-237
• /
• 2013

A newly developed fall detection algorithm using the HMM (Hidden Markov Model) extracted from the video is introduced. To distinguish between the fall from personal difference fall pattern or the normal activities of daily living (ADL), HMM machine learning algorithm is used. For getting fall feature vector of video, the motion vector from the optical flow is applied to the PCA (Principal Component Analysis). The combination of the angle, ratio of long-short axis, velocity from results of PCA make the new fall feature parameters. These parameters were applied to the HMM and the results were compared and analyzed. Among the newly proposed various kinds of fall parameters, the angle of movement showed the best results. The results show that this parameter can distinguish various types of fall from ADLs with 91.5% sensitivity and 88.01% specificity.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2013.10a
• /
• pp.393-395
• /
• 2013

In modern society, the elderly over 65 years old are increasing due to development of medical technology and improvement of their standard of living. Severe fall of the elderly can lead to death threats. To solve this problem, several algorithms and hardware systems for fall detection have been studied and developed. In this paper, a fall detection system using 3-axial acceleration sensor is presented. In the fall detection system, several types of fall-feature parameters are calculated and then the fall is determined by using them. Using this system, best sensitivity and specificity are 98.3% and 94.7%, respectively.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.9
• /
• pp.2101-2108
• /
• 2014

Automatic fall-detection algorithms using video-data are proposed. Six types of fall-feature parameters are defined applying the optical flows extracted from differential images to principal component analysis(PCA). One fall-detection algorithm is the simple threshold method that a fall is detected when a fall-feature parameter is over a threshold, another is to use the HMM, and the other is to combine the simple threshold and HMM. Comparing the performances of three types of fall-detection algorithm, the algorithm combining the simple threshold and HMM requires less computational resources than HMM and exhibits a higher accuracy than the simple threshold method.

• Smart Media Journal
• /
• v.10 no.3
• /
• pp.31-38
• /
• 2021

It is hard to predict when and where a fall accident will happen. Also, if rapid follow-up measures on it are not performed, a fall accident leads to a threat of life, so studies that can automatically detect a fall accident have become necessary. Among automatic fall-accident detection techniques, a fall detection scheme using an IMU (inertial measurement unit) sensor attached to a wrist is difficult to detect a fall accident due to its movement, but it is recognized as a technique that is easy to wear and has excellent accessibility. To overcome the difficulty in obtaining fall data, this study proposes an algorithm that efficiently learns less data through machine learning such as KNN (k-nearest neighbors) and SVM (support vector machine). In addition, to improve the performance of these mathematical classifiers, this study utilized feature data aquired in the frequency space. The proposed algorithm analyzed the effect by diversifying the parameters of the model and the parameters of the frequency feature extractor through experiments using standard datasets. The proposed algorithm could adequately cope with a realistic problem that fall data are difficult to obtain. Because it is lighter than other classifiers, this algorithm was also easy to implement in small embedded systems where SIMD (single instruction multiple data) processing devices were difficult to mount.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.6
• /
• pp.254-259
• /
• 2013

A newly developed fall recognition algorithm using gravity weighted 3-axis accelerometer data as the input of HMM (Hidden Markov Model) is introduced. Five types of fall feature parameters including the sum vector magnitude(SVM) and a newly-defined gravity-weighted sum vector magnitude(GSVM) are applied to a HMM to evaluate the accuracy of fall recognition. A GSVM parameter shows the best accuracy of falls which is 100% of sensitivity and 97.96% of specificity, and comparing with SVM, the results archive more improved recognition rate, 5.2% of sensitivity and 4.5% of specificity. GSVM shows higher recognition rate than SVM due to expressing falls characteristics well, whereas SVM expresses the only momentum.

• Structural Engineering and Mechanics
• /
• v.52 no.6
• /
• pp.1193-1208
• /
• 2014

A simulation of failures on responsible elements is only one form of the extreme structural behavior analysis. By understanding the dynamic behavior in incidental situations, it is possible to make a special structural design from the point of the largest axial force, stress and redundancy. The numerical realization of one such simulation analysis was performed using FEM in this paper. The boundary parameters of transient analysis, such as overall structural damping coefficient, load accelerations, time of load fall and internal forces in the responsible structural elements, were determined on the basis of the dynamic experimental parameters. The structure eigenfrequencies were determined in modal analysis. In the study, the basic incidental models were set. The models were identified by many years of monitoring incidental situations and the most frequent human errors in work with heavy structures. The combined load models of structure are defined in the paper since the incidents simply arise as consequences of cumulative errors and failures. A feature of a combined model is that the single incident causes the next incident (consecutive timing) as well as that other simple dynamic actions are simultaneous. The structure was observed in three typical load positions taken from the crane passport (range-load). The obtained dynamic responses indicate the degree of structural sensitivity depending on the character of incident. The dynamic coefficient KD was adopted as a parameter for the evaluation of structural sensitivity.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2010.02a
• /
• pp.464-464
• /
• 2010

A few years ago, for maintaining high stability and production yield of production equipment in a semiconductor fab, on-line monitoring of wafers is required, so that semiconductor manufacturers are investigating a software based process controlling scheme known as virtual metrology (VM). As semiconductor technology develops, the cost of fabrication tool/facility has reached its budget limit, and reducing metrology cost can obviously help to keep semiconductor manufacturing cost. By virtue of prediction, VM enables wafer-level control (or even down to site level), reduces within-lot variability, and increases process capability, $C_{pk}$. In this research, we have practiced VM on $SiO_2$ etch rate with optical emission spectroscopy(OES) data acquired in-situ while the process parameters are simultaneously correlated. To build process model of $SiO_2$ via, we first performed a series of etch runs according to the statistically designed experiment, called design of experiments (DOE). OES data are automatically logged with etch rate, and some OES spectra that correlated with $SiO_2$ etch rate is selected. Once the feature of OES data is selected, the preprocessed OES spectra is then used for in-situ sensor based VM modeling. ICP-RIE using 葰.56MHz, manufactured by Plasmart, Ltd. is employed in this experiment, and single fiber-optic attached for in-situ OES data acquisition. Before applying statistical feature selection, empirical feature selection of OES data is initially performed in order not to fall in a statistical misleading, which causes from random noise or large variation of insignificantly correlated responses with process itself. The accuracy of the proposed VM is still need to be developed in order to successfully replace the existing metrology, but it is no doubt that VM can support engineering decision of "go or not go" in the consecutive processing step.

• The Korean Journal of Quaternary Research
• /
• v.11 no.1
• /
• pp.57-68
• /
• 1997

Climatic change of the late-Quaternary period has been record-ed in the loess deposits of the central Great plains and the record of such change is extractable using a number of approaches and parameters. The stratigraphy of loess deposits which have been investigated on Fort Riley exhibits the same sequence of loess units and intercalated buried soils as is found elsewhere in the re-gion but adds detail unique to the reservation Upland late-Qua-ternary composite stratigraphy preserved on the reservation con-sists of the basal Sangamon soil of the Last interglacial(c. 120-110ka), Gilman Canyon Formation(c. >40 -20ka), Peoria loess(c. 20 -10ka) Brady soil(c. 11 -10ka) Bignell loess(c. 9-\ulcornerka). and mod-ern surface soil. Application of magnetic analyses has provided proxy data sets that represent a time series of climatically regulated pedogenesis/weathering and botanical composition. magetic data have yielded an impression of the variation in climate from Sangamon time to the late Holocene through a reconstruction of the history of pedogenesis/weathering. Sangamon soil formation dominated the reservation durin the Last interglacial as indicated by magnetic parameters. During Gil-man Canyon time loess influx was usually sufficiently slow as to permit pedogenesis which appears to have been at a maximum twice during that time. Warm season grasses were important dur-ing soil formation but diminished in importance during the peri-ods of more rapid loess fall which were cooler and perhaps wet-ter. Peoria loess fall a function of the deterioration of climate during the last Glacial Maximum thinly blanketed the reservation with thickest accumulations occurring to the north-west(Bala Cemetery site)proximal to the source region. Long-term surface stability did not apparently occur within Peoria time but short-term stability may be indicaed by the presence of thin weathering zones(incipient soils) in the Peoria loess. Re-gional landscape stability prevailed during the environmental shift at the Pleistocene/Holocene transition resulting in forma-tion of the well expressed Brady soil. One or more weak soils developed in the Bignell loess as it ac-cumulated. A notable feature of the Bignell loess is the appear-ance of the Altithermal dry period: the loess experienced little weathering and was dominated by warm season grasses until the latter of the Holocene.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.14 no.5
• /
• pp.969-976
• /
• 2019

A lightweight artificial intelligence hardware has made great strides in many application areas. In general, a lightweight artificial intelligence system consist of lightweight artificial intelligence engine and preprocessor including feature selection, generation, extraction, and normalization. In order to achieve optimal performance in broad range of applications, lightweight artificial intelligence system needs to choose a good preprocessing function and set their respective hyper-parameters. This paper proposes a unified framework for a lightweight artificial intelligence system and utilization method for finding models with optimal performance to use on a given dataset. The proposed unified framework can easily generate a model combined with preprocessing functions and lightweight artificial intelligence engine. In performance evaluation using handwritten image dataset and fall detection dataset measured with inertial sensor, the proposed unified framework showed building optimal artificial intelligence models with over 90% test accuracy.