• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.9
• /
• pp.349-358
• /
• 2021

Recently, a method of finding the attention area or localization area for an object of an image using CAM (Class Activation Map)[1] has been variously carried out as a study of WSOL (Weakly Supervised Object Localization). The attention area extraction from the object heat map using CAM has a disadvantage in that it cannot find the entire area of the object by focusing mainly on the part where the features are most concentrated in the object. To improve this, using CAM and Selective Search[6] together, we first expand the attention area in the heat map, and a Gaussian smoothing is applied to the extended area to generate retraining data. Finally we train the data to expand the attention area of the objects. The proposed method requires retraining only once, and the search time to find an localization area is greatly reduced since the selective search is not needed in this stage. Through the experiment, the attention area was expanded from the existing CAM heat maps, and in the calculation of IOU (Intersection of Union) with the ground truth for the bounding box of the expanded attention area, about 58% was improved compared to the existing CAM.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.4
• /
• pp.474-480
• /
• 2020

With the advent of the Fourth Industrial Revolution, many new technologies are being utilized. In particular, video signals are used in various fields. However, when transmitting and receiving video signals, salt and pepper noise and additive white Gaussian noise (AWGN) occur for multiple reasons. Failure to remove such noise when performing image processing can cause problems. Generally, filters such as CWMF, MF, and AMF remove noise. However, these filters perform somewhat poorly in the high-density noise domain and cause smoothing, resulting in slightly lower retention of the edge components. In this paper, we propose an algorithm by effectively eliminating salt and pepper noise using a modified weight filter using standard deviation. In order to prove the noise reduction performance of the proposed algorithm, we compared it with the existing algorithm using PSNR and magnified images.

• Journal of Intelligence and Information Systems
• /
• v.29 no.2
• /
• pp.115-127
• /
• 2023

Video stabilization is one of the camera technologies that the importance is gradually increasing as the personal media market has recently become huge. For deep learning-based video stabilization, existing methods collect pairs of video datas before and after stabilization, but it takes a lot of time and effort to create synchronized datas. Recently, to solve this problem, unsupervised learning method using only unstable video data has been proposed. In this paper, we propose a network structure that learns the stabilized trajectory only with the unstable video image without the pair of unstable and stable video pair using the Convolutional Auto Encoder structure, one of the unsupervised learning methods. Optical flow data is used as network input and output, and optical flow data was mapped into grid units to simplify the network and minimize noise. In addition, to generate a stabilized trajectory with an unsupervised learning method, we define the loss function that smoothing the input optical flow data. And through comparison of the results, we confirmed that the network is learned as intended by the loss function.

• Progress in Medical Physics
• /
• v.23 no.1
• /
• pp.62-69
• /
• 2012

The dose re-calculation process using Megavoltage cone-beam CT images is inevitable process to perform the Adaptive Radiation Therapy (ART). The purpose of this study is to improve dose re-calculation accuracy using MVCBCT images by applying intensity calibration method and three dimensional rigid body transform and filtering process. The three dimensional rigid body transform and Gaussian smoothing filtering process to MVCBCT Rando phantom images was applied to reduce image orientation error and the noise of the MVCBCT images. Then, to obtain the predefined modification level for intensity calibration, the cheese phantom images from kilo-voltage CT (kV CT), MVCBCT was acquired. From these cheese phantom images, the calibration table for MVCBCT images was defined from the relationship between Hounsfield Units (HUs) of kV CT and MVCBCT images at the same electron density plugs. The intensity of MVCBCT images from Rando phantom was calibrated using the predefined modification level as discussed above to have the intensity of the kV CT images to make the two images have the same intensity range as if they were obtained from the same modality. Finally, the dose calculation using kV CT, MVCBCT with/without intensity calibration was applied using radiation treatment planning system. As a result, the percentage difference of dose distributions between dose calculation based on kVCT and MVCBCT with intensity calibration was reduced comparing to the percentage difference of dose distribution between dose calculation based on kVCT and MVCBCT without intensity calibration. For head and neck, lung images, the percentage difference between kV CT and non-calibrated MVCBCT images was 1.08%, 2.44%, respectively. In summary, our method has quantitatively improved the accuracy of dose calculation and could be a useful solution to enhance the dose calculation accuracy using MVCBCT images.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.45 no.6
• /
• pp.123-131
• /
• 2008

In Oriental medicine, the status of a tongue is the important indicator to diagnose the condition of one's health like the physiological and the clinicopathological changes of internal organs in a body. A tongue diagnosis is not only convenient but also non-invasive, and therefore widely used in Oriental medicine. However, the tongue diagnosis is affected by examination circumstances like a light source, patient's posture, and doctor's condition a lot. To develop an automatic tongue diagnosis system for an objective and standardized diagnosis, segmenting a tongue region from a facial image captured and classifying tongue coating are inevitable but difficult since the colors of a tongue, lips, and skin in a mouth are similar. The proposed method includes preprocessing, over-segmenting, detecting the edge with a local minimum over a shading area from the structure of a tongue, correcting local minima or detecting the edge with the greatest color difference, selecting one edge to correspond to a tongue shape, and smoothing edges, where preprocessing consists of down-sampling to reduce computation time, histogram equalization, and edge enhancement, which produces the region of a segmented tongue. Finally, the systematic procedure separated only a tongue region from a face image with a tongue, which was obtained from a digital tongue diagnosis system. Oriental medical doctors' evaluation for the results illustrated that the segmented region excluding a non-tongue region provides important information for the accurate diagnosis. The proposed method can be used for an objective and standardized diagnosis and for an u-Healthcare system.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.19 no.1
• /
• pp.12-16
• /
• 2015

Purpose Principal component analysis (PCA) is a method often used in the neuroimagre analysis as a multivariate analysis technique for describing the structure of high dimensional correlation as the structure of lower dimensional space. PCA is a statistical procedure that uses an orthogonal transformation to convert a set of observations of correlated variables into a set of values of linearly independent variables called principal components. In this study, in order to investigate the usefulness of PCA in the brain PET image analysis, we tried to analyze C[11]-PIB PET image as a representative case. Materials and Methods Nineteen subjects were included in this study (normal = 9, AD/MCI = 10). For C[11]-PIB, PET scan were acquired for 20 min starting 40 min after intravenous injection of 9.6 MBq/kg C[11]-PIB. All emission recordings were acquired with the Biograph 6 Hi-Rez (Siemens-CTI, Knoxville, TN) in three-dimensional acquisition mode. Transmission map for attenuation-correction was acquired using the CT emission scans (130 kVp, 240 mA). Standardized uptake values (SUVs) of C[11]-PIB calculated from PET/CT. In normal subjects, 3T MRI T1-weighted images were obtained to create a C[11]-PIB template. Spatial normalization and smoothing were conducted as a pre-processing for PCA using SPM8 and PCA was conducted using Matlab2012b. Results Through the PCA, we obtained linearly uncorrelated independent principal component images. Principal component images obtained through the PCA can simplify the variation of whole C[11]-PIB images into several principal components including the variation of neocortex and white matter and the variation of deep brain structure such as pons. Conclusion PCA is useful to analyze and extract the main pattern of C[11]-PIB image. PCA, as a method of multivariate analysis, might be useful for pattern recognition of neuroimages such as FDG-PET or fMRI as well as C[11]-PIB image.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.13 no.1
• /
• pp.35-39
• /
• 2009

Purpose: Diagnostic and functional imaging softwares in Nuclear Medicine have been developed significantly. But, there are some limitations which like take a lot of time. In this article, we introduced that the basic concept of macro to help understanding macro and its application to Brain SPECT processing. We adopted macro software to SPM processing and PACS verify processing of Brain SPECT processing. Materials and Methods: In Brain SPECT, we choose SPM processing and two PACS works which have large portion of a work. SPM is the software package to analyze neuroimaging data. And purpose of SPM is quantitative analysis between groups. Results are made by complicated process such as realignment, normalization, smoothing and mapping. We made this process to be more simple by using macro program. After sending image to PACS, we directly input coordinates of mouse using simple macro program for processes of color mapping, adjustment of gray scale, copy, cut and match. So we compared time for making result by hand with making result by macro program. Finally, we got results by applying times to number of studies in 2007. Results: In 2007, the number of SPM studies were 115 and the number of PACS studies were 834 according to Diamox study. It was taken 10 to 15 minutes for SPM work by hand according to expertness and 5 minutes and a half was uniformly needed using Macro. After applying needed time to the number of studies, we calculated an average time per a year. When using SPM work by hand according to expertness, 1150 to 1725 minutes (19 to 29 hours) were needed and 632 seconds (11 hours) were needed for using Macro. When using PACS work by hand, 2 to 3 minutes were needed and for using Macro, 45 seconds were needed. After applying theses time to the number of studies, when working by hand, 1668 to 2502 minutes (28 to 42 hours) were needed and for using Macro, 625 minutes (10 hours) were needed. Following by these results, it was shown that 1043 to 1877 (17 to 31 hours were saved. Therefore, we could save 45 to 63% for SPM, 62 to 75% for PACS work and 55 to 70% for total brain SPECT processing in 2007. Conclusions: On the basis of the number of studies, there was significant time saved when we applied Macro to brain SPECT processing and also it was shown that even though work is taken a little time, there is a possibility to save lots of time according to the number of studies. It gives time on technologist's side which makes radiological technologist more concentrate for patients and reduce probability of mistake. Appling Macro to brain SPECT processing helps for both of radiological technologists and patients and contribute to improve quality of hospital service.

• Journal of Biomedical Engineering Research
• /
• v.25 no.3
• /
• pp.189-194
• /
• 2004

Pixel values of contrast enhanced computed tomography (CE-CT) images are randomly changed. Also, the middle liver part has a problem to segregate the liver structure because of similar gray-level values of a pancreas in the abdomen. In this paper, an automatic liver segmentation method using a partial histogram threshold (PHT) algorithm is proposed for overcoming randomness of CE-CT images and removing the pancreas. After histogram transformation, adaptive multi-modal threshold is used to find the range of gray-level values of the liver structure. Also, the PHT algorithm is performed for removing the pancreas. Then, morphological filtering is processed for removing of unnecessary objects and smoothing of the boundary. Four CE-CT slices of eight patients were selected to evaluate the proposed method. As the average of normalized average area of the automatic segmented method II (ASM II) using the PHT and manual segmented method (MSM) are 0.1671 and 0.1711, these two method shows very small differences. Also, the average area error rate between the ASM II and MSM is 6.8339 ％. From the results of experiments, the proposed method has similar performance as the MSM by medical Doctor.

• Korean Security Journal
• /
• no.24
• /
• pp.53-66
• /
• 2010

This study aims to analyze kinematic variables regarding lateral break-fall quantitatively that can protect the body, prevent the injury and minimize the impact. To this end, three Hapkido, judo athletes of H University with experience of over 5 years were selected. Test was conducted through three dimensional image analysis by checking the time and order the subjects reach a mat. In this study, lateral break-fall was repeated five times and among them, the best movement was selected. The picture shot with high-speed camera was analyzed by using KWON3D ver. 3.1 program through three dimensional coordinate calculation based on DLT method and smoothing process of data. Study results were as follows. 1. With respect to time variables shown in lateral break-fall of A, B, C athletes, there is small difference in temporal variables and in the order the body reaches a mat. With respect to average value, hand is ($0.94{\pm}0.20$), elbow ($0.97{\pm}0.17$), hip ($0.97{\pm}0.18$), back ($0.98{\pm}0.18$), and shoulder ($1.04{\pm}0.16$). Time variable the body reaches a mat in lateral break-fall is in hand, elbow, hip, back and shoulder. 2. With respect to moving distance variables shown in lateral break-fall of A, B, C athletes, hand is ($34.33{\pm}34.59$), elbow ($52.00{\pm}26.06$), hip ($70.00{\pm}15.72$), back ($153.67{\pm}17.93$), and should ($130.67{\pm}29.02$). The fact that this study contributed to improving security martial arts technique and protecting the body by understanding the principle of lateral break-fall movement is of significance. In addition, the fact that this study provided systematic basic data for improving security martial arts technique is significant.

• Korean Journal of Remote Sensing
• /
• v.31 no.5
• /
• pp.449-459
• /
• 2015

The conventional National Forest Inventory(NFI)-based forest carbon stock estimation method is suitable for national-scale estimation, but is not for regional-scale estimation due to the lack of NFI plots. In this study, for the purpose of regional-scale carbon stock estimation, we created grid-based forest carbon stock maps using spatial ancillary data and two types of up-scaling methods. Chungnam province was chosen to represent the study area and for which the $5^{th}$ NFI (2006~2009) data was collected. The first method (method 1) selects forest type map as ancillary data and uses regression model for forest carbon stock estimation, whereas the second method (method 2) uses satellite imagery and k-Nearest Neighbor(k-NN) algorithm. Additionally, in order to consider uncertainty effects, the final AGB carbon stock maps were generated by performing 200 iterative processes with Monte Carlo simulation. As a result, compared to the NFI-based estimation(21,136,911 tonC), the total carbon stock was over-estimated by method 1(22,948,151 tonC), but was under-estimated by method 2(19,750,315 tonC). In the paired T-test with 186 independent data, the average carbon stock estimation by the NFI-based method was statistically different from method2(p<0.01), but was not different from method1(p>0.01). In particular, by means of Monte Carlo simulation, it was found that the smoothing effect of k-NN algorithm and mis-registration error between NFI plots and satellite image can lead to large uncertainty in carbon stock estimation. Although method 1 was found suitable for carbon stock estimation of forest stands that feature heterogeneous trees in Korea, satellite-based method is still in demand to provide periodic estimates of un-investigated, large forest area. In these respects, future work will focus on spatial and temporal extent of study area and robust carbon stock estimation with various satellite images and estimation methods.