• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.48 no.2
• /
• pp.74-79
• /
• 2011

HDR(high dynamic range) imaging techniques supports wider dynamic range than normal images captured from general still camera. These usually need several shots to obtain LDR(low dynamic range) images, causing ghosting artifacts. Accordingly, this paper suggests a method to generate new LDR images from a single Bayer image using Exposure LUT(look-up table) by considering channel dependency. We prior construct exposure LUT for each RGB channel, showing the relationship between input and average output luminance values. In the process, by applying the average luminance of input image and current exposure to LUT, new exposures which are determined by user choice are first estimated. Next, LDR images which are corresponded to new exposures are generated based on each LUT. Saturated areas are improved by considering channel dependency in the last procedure. In the experimental comparison, high PSNR values are obtained between estimated and captured images. Also, we have similar appearance on displayed images.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.49 no.4
• /
• pp.58-66
• /
• 2012

Currently many devices reproduce electronic images in the various ways. However, the color that is reproduced in any device is different from the original color due to the differences in the gamut between devices. A recent trend in gamut mapping algorithms is the use of spatial information to compute the color transformation of pixels from the input to the output gamut. However, these techniques share the problem of preserving details, and avoiding halos, and hue shift. In this paper, spatial gamut mapping for preserving the details of the input image is proposed. Our approach improves visibility of detail that is not effectively represented with conventional spatial gamut mapping. In proposed method, initially, we gamut map the input image using gamut clipping and obtain a detail layer for both the input and the gamut mapped images. Next, we calculate the difference between the two detail layers, obtaining the details of the out of gamut region. Finally, we add the details of out of gamut region to the gamut mapped image. Since the resulting image has out of gamut colors, we obtain resulting image of proposed method by using a gamut clipping method. Consequently, the printed output image was more consistent with the corresponding monitor image.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.12 no.6
• /
• pp.629-635
• /
• 2019

In Deep Learning models derivative is implemented by error back-propagation which enables the model to learn the error and update parameters. It can find the global (or local) optimal points of parameters even in the complex models taking advantage of a huge improvement in computing power. However, deliberately generated data points can 'fool' models and degrade the performance such as prediction accuracy. Not only these adversarial examples reduce the performance but also these examples are not easily detectable with human's eyes. In this work, we propose the method to detect adversarial datasets with random noise addition. We exploit the fact that when random noise is added, prediction accuracy of non-adversarial dataset remains almost unchanged, but that of adversarial dataset changes. We set attack methods (FGSM, Saliency Map) and noise level (0-19 with max pixel value 255) as independent variables and difference of prediction accuracy when noise was added as dependent variable in a simulation experiment. We have succeeded in extracting the threshold that separates non-adversarial and adversarial dataset. We detected the adversarial dataset using this threshold.

• The Korean Society of Law and Medicine
• /
• v.17 no.1
• /
• pp.281-297
• /
• 2016

In order to make a treatment plan and outcome prediction, it is important to evaluate accurately and objectively osseous tissues of the implant area. The evaluation of osseous tissues is the most objective method for the decision of production time of upper structure of alveolar bone. However, the evaluation of osseous tissues contains contradiction because it is made by subjective opinions of dental surgeons. Many dentists also point out the problem of subjective evaluation of osseous tissues. Therefore, it is necessary to create accurate and objective standards. Previously, the evaluation of bone density depends on dentist's subjective sensation during drilling procedure of implant. However, the HU(Hounsfield unit) figure of CT(computed tomography) scan allows of objective and precise categorization of bone density now. Misch and Kircos divided the bone density levels from D1 to D5 with subjective separation of bone density. Their method also depended on not objective and quantification data but subjective separation by sensation. Thus, we need the evaluation of implant area through comparative analysis of more objective and quantification data. Implant treatment comprises the highest frequency of medical disputes of dental clinic. If we bring objective checkup and reasonable treatment method in the implant treatment, we can deduce more reasonable results, and the failure late of implant treatment also can decrease. The ultimate objective of this study is the minimization of dental disputes between dental patients and dentists by creating new legal standards on the basis of objective and quantification data.

• Journal of the Korean Society of Radiology
• /
• v.1 no.1
• /
• pp.11-16
• /
• 2007

In this study, we have performed contrast-detail analysis for an amorphous selenium(a-Se) based digital X-ray imaging system by using a contrast-detail phantom(CDRAD 2.0) to test its low contrast performance. The X-ray imaging system utilizes an 500-mm-thick a-Se semiconductor X-ray absorber coated over an amorphous silicon(a-Si) TFT(thin-film transistor) detector matrix with a $139mm{\times}139mm$ pixel size and a $46.7cm{\times}46.7cm$ active area. In the measurement of contrast-detail curves we first acquired X-ray images of the CDRAD 2.0 phantom at given test conditions(i.e., 40, 50, 60, 70, 80 kVp, and 16 mA.s), and then evaluated the contrast-detail characteristics of the imaging system from each phantom image by using an image quality factor called the image-quality-figure-inverse(IQFinv). The IQFinv values for the imaging system gradually improved with the photon fluence, indicating the improvement of image visibility: 24.4, 35.3, 39.2, 41.5, and 43.4 at photon fluences of $1.8{\times}105$, $5.9{\times}105$, $11.3{\times}105$, $19.4{\times}105$, and $29.4{\times}105$ photons/$mm^2$, respectively.

• Journal of the Korea Society of Computer and Information
• /
• v.17 no.3
• /
• pp.51-57
• /
• 2012

Hepatic vessel tree is the key structure for hepatic disease diagnosis and liver surgery planning. Especially, it is used to evaluate the donors' and recipients' liver for the LDLT(Living Donors Liver Transplantation) and estimate the volumes of left and right hepatic lobes for securing their life in the LDLT. In this study, we propose a method to apply canny edge detection that is not affected by noise to the liver images for automatic segmentation of hepatic vessels tree in contrast abdominal MDCT image. Using histograms and average pixel values of the various liver CT images, optimized parameters of the Canny algorithm are determined. It is more time-efficient to use the common parameters than to change parameters manually according to CT images. Candidates of hepatic vessels are extracted by threshold filtering around the detected the vessel edge. Finally, using a system which detects the true-negatives and the false-positives in horizontal and vertical direction, the true-negatives are added in candidate of hepatic vessels and the false-positives are removed. As a result of the process, the various hepatic vessel trees of patients are accurately reconstructed in 3D.

• Korean Journal of Optics and Photonics
• /
• v.22 no.1
• /
• pp.23-29
• /
• 2011

Bistable chiral splay nematic liquid crystal display (BCSN LCD) is a memory type liquid crystal display using splay and $-\pi$ twist states as two stable states. When the cell thickness to pitch (d/p) ratio is 0.25, splay and $-\pi$ twist states have permanent memory time. However, when the transition from $-\pi$ twist state to splay state is caused by a fringe field, pixel regions show that the splay state is not perfect, but rather includes a contribution from the $-\pi$ twist state. In this paper, we propose a reflective BCSN LCD using $-\pi$ twist state in which the two stable states do not coexist. The fabricated reflective BCSN LC cell shows a high contrast ratio of over 30:1 and response times of 950 ms and 450 ms in vertical and fringe field switching, respectively. The proposed cell also shows wide viewing angle characteristics of $180^{\circ}$ in left- and right directions.

• Journal of the Korean Wood Science and Technology
• /
• v.43 no.1
• /
• pp.43-51
• /
• 2015

Wood shrinkage, an important study subject with regard to the use of wood, has long been studied by researchers. However, when the size of a wood specimen is measured, distortion must be taken into account, which can be accomplished by applying external force on the wood specimen. However, when measuring a large number of specimens, this technique can be a lengthy process. If the size is measured and the shrinkage is calculated from images acquired with a flatbed scanner, it is possible to reduce the error in the measurement and to shorten the measurement time because the images of many specimens can be acquired with one scan. To clearly establish the boundary between a wood specimen and the background in a scan, an image threshold method was applied here. The size of a wood specimen measured by means of a scanner image was found to be longer than the value determined with a vernier caliper. The maximum pixel size of a scan image for highly accurate shrinkage calculations compared with the use of a vernier caliper was 0.053 mm/pixel.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.42 no.5 s.305
• /
• pp.173-180
• /
• 2005

In this paper, we proposed the image hiding method which decreases calculation amount by encrypt partial data using discrete wavelet transform(DWT) and linear scale quantization which were adopted as the main technique for frequency transform in JPEG2000 standard. Also we used the chaotic system and cat map which has smaller calculation amount than other encryption algorithms and then dramatically decreased calculation amount. This method operates encryption process between quantization and entropy coding for preserving compression ratio of images and uses the subband selection method. Also, suggested encryption method to JPEG2000 progressive transmission. The experiments have been performed with the Proposed methods implemented in software for about 500 images. Consequently, we are sure that the proposed is efficient image encryption methods to acquire the high encryption effect with small amount of encryption. It has been shown that there exits a relation of trade-off between the execution time and the effect of the encryption. It means that the proposed methods can be selectively used according to the application areas.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.44 no.5
• /
• pp.1-10
• /
• 2007

Most of natural images have a special property, what is called self-similarity, which is the basis of fractal image coding. Even though an image has local stationarity in several homogeneous regions, it is generally non-stationarysignal, especially in edge region. This is the main reason that poor results are induced in linear techniques. In order to overcome the difficulty we propose a non-linear technique using self-similarity in the image. In our work, an image is classified into stationary and non-stationary region with respect to sample variance. In case of stationary region, do-noising is performed as simply averaging of its neighborhoods. However, if the region is non-stationary region, stationalization is conducted as make a set of center pixels by similarity matching with respect to bMSE(block Mean Square Error). And then do-nosing is performed by Gaussian weighted averaging of center pixels of similar blocks, because the set of center pixels of similar blocks can be regarded as nearly stationary. The true image value is estimated by weighted average of the elements of the set. The experimental results show that our method has better performance and smaller variance than other methods as estimator.