• Korean Journal of Digital Imaging in Medicine
• /
• v.9 no.2
• /
• pp.31-38
• /
• 2007

To evaluate the accuracy and extent in the localization of cerebral motor coutex activation using a gradient- echo echo planar imaging(GE-EPI) compared to spin-echo echo planar iimaging(SE-EPI) on 3T MR imaging. Functional MR imaging of cerebral motor cortex activation was examined in GE-EPI and SE-EPI in five healthy male volunteers. A right finger movement was accomplished with a paradigm of 6 task and rest, periods and the cross-correlation was used for a statistical mapping algorithm. We evaluated any sorts of differenced of the time seried and the signal intensity changes between the rest and task periods obtained with two technoques. The qualitative analysis was distributed with activation sites of large veins and small veins by using two techniques and was found that both the techniques were clinically uesful for delineating large veins and small veins in fMRL Signal intensity charge of the rest and activation periods provided simmilar activations in both methods(GE-EPI : 0.93$\pm$0.11, SE-EPI : 0.80$\pm$.015) but the signal intensity in GE-EPI(133.95$\pm$15.76) was larger than in SE-EPI(74.5$\pm$18.90). The average SNRs of EPI raw data were higher at SMA in SE-EPI(48.54$\pm$12.37) than GE-EPI(41.4$\pm$12.54) and at M1 in SE-EPI(43.24$\pm$11.77) than GE-EPI(38.27$\pm$6.53). The localization of activation voxels of the GE-EPI showed a larger vein but the SE-EPI generally showed small vein. Then the analysis results of the two techniques were used for a statistacal paired student t-test. SE-EPI was found clinically useful for localizing the cerebral moter cortex cativation on 3.0T, but showed a little different activation patterns comparad to GE-EPI. In conclusion, SE-EPI may be feasible and can detect true cortical activation from capillaries and GE-EPI can obtain the large veins in the motor cortex activation on 3T MR imaging.

• Investigative Magnetic Resonance Imaging
• /
• v.18 no.4
• /
• pp.290-302
• /
• 2014

Purpose : Susceptibility-weighted magnetic resonance (MR) sequence is three-dimensional (3D), spoiled gradient-echo pulse sequences that provide a high sensitivity for the detection of blood degradation products, calcifications, and iron deposits. This pictorial review is aimed at illustrating and discussing its main clinical applications. Materials and Methods: SWI is based on high-resolution, 3D, fully velocity-compensated gradient-echo sequences using both magnitude and phase images. To enhance the visibility of the venous structures, the magnitude images are multiplied with a phase mask generated from the filtered phase data, which are displayed at best after post-processing of the 3D dataset with the minimal intensity projection algorithm. A total of 200 patients underwent MR examinations that included SWI on a 3 tesla MR imager were enrolled. Results: SWI is very useful in detecting multiple brain disorders. Among the 200 patients, 80 showed developmental venous anomaly, 22 showed cavernous malformation, 12 showed calcifications in various conditions, 21 showed cerebrovascular accident with susceptibility vessel sign or microbleeds, 52 showed brain tumors, 2 showed diffuse axonal injury, 3 showed arteriovenous malformation, 5 showed dural arteriovenous fistula, 1 showed moyamoya disease, and 2 showed Parkinson's disease. Conclusion: SWI is useful in detecting occult low flow vascular lesions, calcification and microbleed and characterising diverse brain disorders.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.2
• /
• pp.31-41
• /
• 2024

Deep learning models show excellent performance in tasks such as image classification and object detection in the field of computer vision, and are used in various ways in actual industrial sites. Recently, research on improving robustness has been actively conducted, along with pointing out that this deep learning model is vulnerable to hostile examples. A hostile example is an image in which small noise is added to induce misclassification, and can pose a significant threat when applying a deep learning model to a real environment. In this paper, we tried to confirm the robustness of the edge-learning classification model and the performance of the adversarial example detection model using it for adversarial examples of various algorithms. As a result of robustness experiments, the basic classification model showed about 17% accuracy for the FGSM algorithm, while the edge-learning models maintained accuracy in the 60-70% range, and the basic classification model showed accuracy in the 0-1% range for the PGD/DeepFool/CW algorithm, while the edge-learning models maintained accuracy in 80-90%. As a result of the adversarial example detection experiment, a high detection rate of 91-95% was confirmed for all algorithms of FGSM/PGD/DeepFool/CW. By presenting the possibility of defending against various hostile algorithms through this study, it is expected to improve the safety and reliability of deep learning models in various industries using computer vision.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.48 no.2
• /
• pp.47-54
• /
• 2011

This paper describes the method to reduce the time-cost for depth estimation of an image by learning, on the basis of the Genetic Algorithm, the image's features. The depth information is estimated from the relationship among features such as the energy value of an image and the gradient of the texture etc. The estimation-time increases due to the large dimension of an image's features used in the estimating process. And the use of the features without consideration of their importance can have an adverse effect on the performance. So, it is necessary to reduce the dimension of an image's features based on the significance of each feature. Evaluation of the method proposed in this paper using benchmark data provided by Stanford University found that the time-cost for feature extraction and depth estimation improved by 60% and the accuracy was increased by 0.4% on average and up to 2.5%.

• Journal of the Korean Data and Information Science Society
• /
• v.8 no.2
• /
• pp.163-171
• /
• 1997

Function approximation from a set of input-output pairs has numerous applications in scientific and engineering areas. Multilayer feedforward neural networks have been proposed as a good approximator of nonlinear function. The back propagation(BP) algorithm allows multilayer feedforward neural networks to learn input-output mappings from training samples. It iteratively adjusts the network parameters(weights) to minimize the sum of squared approximation errors using a gradient descent technique. However, the mapping acquired through the BP algorithm may be corrupt when errorneous training data we employed. When errorneous traning data are employed, the learned mapping can oscillate badly between data points. In this paper we propose a robust BP learning algorithm that is resistant to the errorneous data and is capable of rejecting gross errors during the approximation process, that is stable under small noise perturbation and robust against gross errors.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.5 no.3
• /
• pp.405-417
• /
• 2001

In this paper, since we currently using surveillance system of analog type bring about waste of resource and efficiency deterioration problems, we describe new solution that design and implementation to the digital surveillance system of new type applying compression techniques and encoding techniques of image data using MPEG-2 international standard. Also, we proposed fast motion estimation algorithm requires much less than the convectional digital surveillance camera system. In this paper a fast motion estimation algorithm is proposed the MPEG-2 video encoding. This algorithm is based on a hybrid use of the block matching technique and gradient technique. Also, we describe a method of moving object extraction directly using MPEG-2 video data. Since proposed method is very simple and requires much less computational power than the conventional object detection methods. In this paper we don't use specific H/W and this system is possible only software encoding, decoding and transmission real-time for image data.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.16 no.7
• /
• pp.1495-1504
• /
• 2012

This paper proposes an algorithm to classify and analyze facial features such as eyebrow, eye, mouth and chin based on the geometric features of the face. As a preprocessing process to classify and analyze the facial features, the algorithm extracts the facial features such as eyebrow, eye, nose, mouth and chin. From the extracted facial features, it detects the shape and form information and the ratio of distance between the features and formulated them to evaluation functions to classify 12 eyebrows types, 3 eyes types, 9 mouth types and 4 chine types. Using these facial features, it analyzes a face. The face analysis algorithm contains the information about pixel distribution and gradient of each feature. In other words, the algorithm analyzes a face by comparing such information about the features.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.18 no.2
• /
• pp.206-211
• /
• 2008

Trainee can enhance his perception of and interaction with the real world by displayed virtual objects in simulation using image processing technology. Therefore, it is essential for realistic simulation to determine the occlusion areas of the virtual object produces after registering real image and virtual object exactly. In this paper, we proposed the new method to solve occlusions which happens during virtual target moves according to the simulated route on real image using improved object contour extraction by neighboring edge search and picking algorithm. After we acquire the detailed contour of complex objects by proposed contour extraction algorithm, we extract the three dimensional information of the position happening occlusion by using MER for performance improvement. In the experiment, we compared proposed method with existed method and preyed the effectiveness in the environment which a partial occlusions happens.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.465-468
• /
• 2022

The incidence of traffic accidents is increasing every year, and Korea is among the top OECD countries. In order to improve this, various road traffic laws are being implemented, and various traffic control methods using equipment such as unmanned speed cameras and traffic control cameras are being applied. However, as drivers avoid crackdowns by detecting the location of traffic control cameras in advance through navigation, a mobile crackdown system that can be cracked down is needed, and research is needed to increase the recognition rate of vehicle license plates on the side of the road for accurate crackdown. This paper proposes a method to improve the vehicle number recognition rate on the road side by applying a gradient correction algorithm using image processing. In addition, custom data learning was conducted using a CNN-based YOLO algorithm to improve character recognition accuracy. It is expected that the algorithm can be used for mobile traffic control cameras without restrictions on the installation location.

• Journal of Institute of Control, Robotics and Systems
• /
• v.3 no.2
• /
• pp.169-178
• /
• 1997

This paper presents a new solution approach to moving obstacle avoidance problem of a robot. A new concept, avoidability measure(AVM) is defined to describe the state of a pair of a robot and an obstacle regarding the collision between them. As an AVM, virtual distance function(VDF) is derived as a function of three state variables: the distance from the obstacle to the robot, outward speed of the obstacle relative to the robot, and outward speed of the robot relative to the obstacle. By keeping the virtual distance above some positive limit value, the robot avoids the obstacle. In terms of the VDF, an artificial potential is constructed to repel the robot away from the obstacle and to attract the robot toward a goal location. At every sampling time, the artificial potential field is updated and the force driving the robot is derived from the gradient of the artificial potential field. The suggested algorithm drives the robot to avoid a moving obstacle in real time. Since the algorithm considers the mobility of the obstacle and robot as well as the distance, it is effective for moving obstacle avoidance. Some simulation studies show the effectiveness of the proposed approach.