• The KIPS Transactions:PartB
• /
• v.17B no.5
• /
• pp.345-354
• /
• 2010

Visual tracking techniques for Augmented Reality are classified as either a marker tracking approach or a natural feature tracking approach. Marker-based tracking algorithms can be efficiently implemented sufficient to work in real-time on mobile devices. On the other hand, natural feature tracking methods require a lot of computationally expensive procedures. Most previous natural feature tracking methods include heavy feature extraction and pattern matching procedures for each of the input image frame. It is difficult to implement real-time augmented reality applications including the capability of natural feature tracking on low performance devices. The required computational time cost is also in proportion to the number of patterns to be matched. To speed up the natural feature tracking process, we propose a novel fast tracking method based on optical flow. We implemented the proposed method on mobile devices to run in real-time and be appropriately used with mobile augmented reality applications. Moreover, during tracking, we keep up the total number of feature points by inserting new feature points proportional to the number of vanished feature points. Experimental results showed that the proposed method reduces the computational cost and also stabilizes the camera pose estimation results.

• Korean Journal of Remote Sensing
• /
• v.36 no.3
• /
• pp.475-486
• /
• 2020

Climate change scenarios are the basis of research to cope with climate change, and consist of large-scale spatio-temporal data. From the data point of view, one scenario has a large capacity of about 83 gigabytes or more, and the data format is semi-structured, making it difficult to utilize the data through means such as search, extraction, archiving and analysis. In this study, a tool for analyzing extreme climate events based on spatial information is developed to improve the usability of large-scale, multi-period climate change scenarios. In addition, a pilot analysis is conducted on the time and space in which the heavy rain thresholds that occurred in the past can occur in the future, by applying the developed tool to the RCP8.5 climate change scenario. As a result, the days with a cumulative rainfall of more than 587.6 mm over three days would account for about 76 days in the 2080s, and localized heavy rains would occur. The developed analysis tool was designed to facilitate the entire process from the initial setting through to deriving analysis results on a single platform, and enabled the results of the analysis to be implemented in various formats without using specific commercial software: web document format (HTML), image (PNG), climate change scenario (ESR), statistics (XLS). Therefore, the utilization of this analysis tool is considered to be useful for determining future prospects for climate change or vulnerability assessment, etc., and it is expected to be used to develop an analysis tool for climate change scenarios based on climate change reports to be presented in the future.

• Journal of KIISE:Software and Applications
• /
• v.34 no.12
• /
• pp.1045-1055
• /
• 2007

Latent topic models are statistical models which automatically captures salient patterns or correlation among features underlying a data collection in a probabilistic way. They are gaining an increased popularity as an effective tool in the application of automatic semantic feature extraction from text corpus, multimedia data analysis including image data, and bioinformatics. Among the important issues for the effectiveness in the application of latent topic models to the massive data set is the efficient learning of the model. The paper proposes an accelerated learning technique for PLSA model, one of the popular latent topic models, by an incremental EM algorithm instead of conventional EM algorithm. The incremental EM algorithm can be characterized by the employment of a series of partial E-steps that are performed on the corresponding subsets of the entire data collection, unlike in the conventional EM algorithm where one batch E-step is done for the whole data set. By the replacement of a single batch E-M step with a series of partial E-steps and M-steps, the inference result for the previous data subset can be directly reflected to the next inference process, which can enhance the learning speed for the entire data set. The algorithm is advantageous also in that it is guaranteed to converge to a local maximum solution and can be easily implemented just with slight modification of the existing algorithm based on the conventional EM. We present the basic application of the incremental EM algorithm to the learning of PLSA and empirically evaluate the acceleration performance with several possible data partitioning methods for the practical application. The experimental results on a real-world news data set show that the proposed approach can accomplish a meaningful enhancement of the convergence rate in the learning of latent topic model. Additionally, we present an interesting result which supports a possible synergistic effect of the combination of incremental EM algorithm with parallel computing.

• 한국정보디스플레이학회:학술대회논문집
• /
• 2008.10a
• /
• pp.993-994
• /
• 2008

Electrochromic (EC) devices are capable of reversibly changing their optical properties upon charge injection and extraction induced by the external voltage. The characteristics of the EC device, such as low power consumption, high coloration efficiency, and memory effects under open circuit status, make them suitable for use in a variety of applications including smart windows and electronic papers. Coloration due to reduction or oxidation of redox chromophores can be used for EC devices (e-paper), but the switching time is slow (second level). Recently, with increasing demand for the low cost, lightweight flat panel display with paper-like readability (electronic paper), an EC display technology based on dye-modified $TiO_2$ nanoparticle electrode was developed. A well known organic dye molecule, viologen, was adsorbed on the surface of a mesoporous $TiO_2$ nanoparticle film to form the EC electrode. On the other hand, ZnO is a wide bandgap II-VI semiconductor which has been applied in many fields such as UV lasers, field effect transistors and transparent conductors. The bandgap of the bulk ZnO is about 3.37 eV, which is close to that of the $TiO_2$ (3.4 eV). As a traditional transparent conductor, ZnO has excellent electron transport properties, even in ZnO nanoparticle films. In the past few years, one-dimension (1D) nanostructures of ZnO have attracted extensive research interest. In particular, 1D ZnO nanowires renders much better electron transportation capability by providing a direct conduction path for electron transport and greatly reducing the number of grain boundaries. These unique advantages make ZnO nanowires a promising matrix electrode for EC dye molecule loading. ZnO nanowires grow vertically from the substrate and form a dense array (Fig. 1). The ZnO nanowires show regular hexagonal cross section and the average diameter of the ZnO nanowires is about 100 nm. The cross-section image of the ZnO nanowires array (Fig. 1) indicates that the length of the ZnO nanowires is about $6\;{\mu}m$. From one on/off cycle of the ZnO EC cell (Fig. 2). We can see that, the switching time of a ZnO nanowire electrode EC cell with an active area of $1\;{\times}\;1\;cm^2$ is 170 ms and 142 ms for coloration and bleaching, respectively. The coloration and bleaching time is faster compared to the $TiO_2$ mesoporous EC devices with both coloration and bleaching time of about 250 ms for a device with an active area of $2.5\;cm^2$. With further optimization, it is possible that the response time can reach ten(s) of millisecond, i.e. capable of displaying video. Fig. 3 shows a prototype with two different transmittance states. It can be seen that good contrast was obtained. The retention was at least a few hours for these prototypes. Being an oxide, ZnO is oxidation resistant, i.e. it is more durable for field emission cathode. ZnO nanotetropods were also applied to realize the first prototype triode field emission device, making use of scattered surface-conduction electrons for field emission (Fig. 4). The device has a high efficiency (field emitted electron to total electron ratio) of about 60%. With this high efficiency, we were able to fabricate some prototype displays (Fig. 5 showing some alphanumerical symbols). ZnO tetrapods have four legs, which guarantees that there is one leg always pointing upward, even using screen printing method to fabricate the cathode.

• The KIPS Transactions:PartB
• /
• v.11B no.3
• /
• pp.281-288
• /
• 2004

In order to extract and analyze complex features of the behavior of animals in response to external stimuli such as toxic chemicals, we implemented an adaptive computational method to characterize changes in the behavior of chironomids in response to treatment with the insecticide, diazinon. In this paper, we propose an energy minimization model to extract the features of response behavior of chironomids under toxic treatment, which is applied on the image of velocity vectors. It is based on the improved active contour model and the variations of the energy functional, which are produced by the evolving active contour. The movement tracks of individual chironomid larvae were continuously measured in 0.25 second intervals during the survey period of 4 days before and after the treatment. Velocity on each sample track at 0.25 second intervals was collected in 15-20 minute periods and was subsequently checked to effectively reveal behavioral states of the specimens tested. Active contour was formed around each collection of velocities to gradually evolve to find the optimal boundaries of velocity collections through processes of energy minimization. The active contour which is improved by T. Chan and L. Vese is used in this paper. The energy minimization model effectively revealed characteristic patterns of behavior for the treatment versus no treatment, and identified changes in behavioral states .is the time progressed.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.6
• /
• pp.228-237
• /
• 2013

In this paper, we propose a normalized cross correlation(NCC)-based multiview background subtraction method which is robust when an object and background have similar color. When the background of the capturing environment is not artificially composed, the regions in the background images which would be occluded by an object tends to have difference colors. The colors of those regions, however, becomes similar when an object enters the capturing environment. Based on this assumption, this paper proposes a concept of GoNCC(Graph of Normalized Cross Correlation). GoNCC is the distribution of NCC between a pixel in an image and pixels related by epipolar constraints with the pixel. The proposed multiview background subtraction method is performed by comparing GoNCC of the current images with the background images. To reduce computational complexity, we perform multiview background subtraction only to the pixels undetermined by single view background subtraction. Experimental results show that the proposed method is more robust to color similarity between an object and background than a single-view background subtraction method and a previous multiview background subtraction method.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.2
• /
• pp.69-80
• /
• 1999

Recently, social requirements of personal identification techniques are rapidly expanding in a number of new application ares. Especially fingerprint recognition is the most important technology. Fingerprint recognition technologies are well established, proven, cost and legally accepted. Therefore, it has more spot lighted among the any other biometrics technologies. In this paper we propose a new on-line fingerprint recognition algorithm for non-inked type live scanner to fit their increasing of security level under the computing environment. Fingerprint recognition system consists of two distinct structural blocks: feature extraction and feature matching. The main topic in this paper focuses on the feature matching using the fingerprint minutiae (ridge ending and bifurcation). Minutiae matching is composed in the alignment stage and matching stage. Success of optimizing the alignment stage is the key of real-time (on-line) fingerprint recognition. Proposed alignment algorithm using clique shows the strength in the search space optimization and partially incomplete image. We make our own database to get the generality. Using the traditional statistical discriminant analysis, 0.05% false acceptance rate (FAR) at 8.83% false rejection rate (FRR) in 1.55 second average matching speed on a Pentium system have been achieved. This makes it possible to construct high performance fingerprint recognition system.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.1
• /
• pp.116-127
• /
• 1996

In this paper, a scheme of classification of scaled textured images using normalized pattern spectrum incorporating arbitrary scale changes based on mathematical morphology is proposed in more general environments considering camera's zoom-in and zoom-out function. The normalized pattern spectrum means that firstly pattern spectrum is calculated and secondly interpolation is performed to incorporate scale changes according to scale change ratio in the same textured image class. Pattern spectrum is efficiently obtained by using both opening and closing, that is, we calculate pattern spectrum by opening method for pixels which have value more than threshold and calculate pattern spectrum by closing method for pixels which have value less than threshold. Also we compare classification accuracy between gray scale method and binary method. The proposed approach has the advantage of efficient information extraction, high accuracy, less computation, and parallel implementation. An important advantage of the proposed method is that it is possible to obtain high classification accuracy with only (1:1) scale images for training phase.

• Journal of the Korean Geographical Society
• /
• v.39 no.5 s.104
• /
• pp.786-803
• /
• 2004

These days, interests on land cover classification using not only multi-sensor data but also thematic GIS information, are increasing. Often, although we have useful GIS information for the classification, the traditional classification method like maximum likelihood estimation technique (MLE) does not allow us to use the information due to the fact that the MLE and the existing computer programs cannot handle GIS data properly. We proposed a new method for updating the image classification using multi-spectral and multi-temporal images. In this study, we have simultaneously extended the MLE to accommodate both multi-spectral images data and land cover data for land cover classification. In addition to the extended MLE method, we also have extended the empirical likelihood ratio estimation technique (LRE), which is one of non-parametric techniques, to handle simultaneously both multi-spectral images data and land cover data. The proposed procedures were evaluated using land cover map based on Landsat ETM+ images in the Anmyeon-do area in South Korea. As a result, the proposed methods showed considerable improvements in classification accuracy when compared with other single-spectral data. Improved classification images showed that the overall accuracy indicated an improvement in classification accuracy of $6.2\%$ when using MLE, and $9.2\%$ for the LRE, respectively. The case study also showed that the proposed methods enable the extraction of the area with land cover change. In conclusion, land cover classification produced through the combination of various GIS spatial data and multi-spectral images will be useful to involve complementary data to make more accurate decisions.

• Journal of Broadcast Engineering
• /
• v.13 no.4
• /
• pp.427-439
• /
• 2008

This paper proposes a novel method to convert monoscopic soccer videos to stereoscopic videos. Through the soccer video analysis process, we detect shot boundaries and classify soccer frames into long shot or non-long shot. In the long shot case, the depth mapis generated relying on the size of the extracted ground region. For the non-long shot case, the shot is further partitioned into three types by considering the number of ground blocks and skin blocks which is obtained by a simple skin-color detection method. Then three different depth assignment methods are applied to each non-long shot types: 1) Depth estimation by object region extraction, 2) Foreground estimation by using the skin block and depth value computation by Gaussian function, and 3)the depth map generation for shots not containing the skin blocks. This depth assignment is followed by stereoscopic image generation. Subjective evaluation comparing generated depth maps and corresponding stereoscopic images indicate that the proposed algorithm can yield the sense of depth from a single view images.