• ETRI Journal
• /
• v.32 no.6
• /
• pp.871-880
• /
• 2010

Many applications dealing with image management need a technique for removing duplicate images or for grouping related (near-duplicate) images in a database. This paper proposes a concentric circle-based image signature which makes it possible to detect near-duplicates rapidly and accurately. An image is partitioned by radius and angle levels from the center of the image. Feature values are calculated using the average or variation between the partitioned sub-regions. The feature values distributed in sequence are formed into an image signature by hash generation. The hashing facilitates storage space reduction and fast matching. The performance was evaluated through discriminability and robustness tests. Using these tests, the particularity among the different images and the invariability among the modified images are verified, respectively. In addition, we also measured the discriminability and robustness by the distribution analysis of the hashed bits. The proposed method is robust to various modifications, as shown by its average detection rate of 98.99%. The experimental results showed that the proposed method is suitable for near-duplicate detection in large databases.

• Proceedings of the KSMRM Conference
• /
• 2001.11a
• /
• pp.112-112
• /
• 2001

Purpose： We developed the software for diffusion tensor imaging and evaluated its feasibility in norm brains. Method： Five normal volunteers, aged from 25 to 29 years, were examined on a 1.5 T MR system. the diffusion tensor pulse sequence used a SE-EPI with 6 diffusion gradie directions of (1, 1, 0), (-1, 1,0), (1, 0, 1), (-1, 0, 1), (0, 1, 1), (0, 1, -1) and also with no diffusion gradient. A b-factor of 500 sec/mm2 was used. Measurement parameter were as follows； TR/TE=10000 ms/99 ms, FOV=240 mm, matrix=128$\times$128, slice thickness/gap=6 mm/0 mm, bandwidth=91 kHz and the number of total slices=20. Four repeated axial diffusion images were averaged for diffusion tensor imaging. A total scan 11 of 4 min 30 sec was used. Six full diffusion tensor components of Dxx, Dyy, Dzz, Dxy, Dxz and Dyz were obtained using two-point linear regression model from 7 diffusion-weight images at each pixel and fractional anisotropy and lattice index images was estimated fr their eigenvectors and eigenvalues. Our program was written on a platform of IDL. W evaluated the qualities of fractional anisotropy and lattice index images of normal brains a knew whether our software for diffusion tensor imaging may be feasible.

• Journal of Korea Multimedia Society
• /
• v.5 no.5
• /
• pp.494-504
• /
• 2002

Volume rendering is a method of displaying volumetric data as a sequence two-dimensional image. Because this algorithm has an advantage of visualizing structures within objects, it has recently been used to analyze medical images i.e, MRI, PET, and SPECT. In this paper. we suggested a method for creating images easily from sampled volumetric data and applied the interpolation method to medical images. Additionally, we implemented and applied two kinds of interpolation methods to improve the image quality, linear interpolation and cubic interpolation at the sampling stage. Subsequently, we compared the results of volume rendered data using a transfer function. We anticipate a significant contribution to diagnosis through image reconstruction using a volumetric data set, because volume rendering techniques of medical images are the result of 3-dimensional data.

• The KIPS Transactions:PartB
• /
• v.8B no.6
• /
• pp.705-717
• /
• 2001

In this paper, we propose a rapid lane detection method to extract a preceding vehicle from sequential images captured by a single monocular CCD camera. We detect positions of lanes for an individual image within the limited area that would not be hidden and thereby compute the slopes of the detected lanes. Then we find a search area where vehicles would exist and extract the position of the preceding vehicle within the area with edge component by applying a structured method. To verify the effects of the proposed method, we capture the road images with a notebook PC and a CCD camera for PC and present the results such as processing time for lane detection, accuracy and vehicles detection against the images.

• Smart Structures and Systems
• /
• v.25 no.6
• /
• pp.733-749
• /
• 2020

This study presents a computer vision-based approach for representing time evolution of structural damages leveraging a database of inspection images. Spatially incoherent but temporally sorted archival images captured by robotic cameras are exploited to represent the damage evolution over a long period of time. An access to a sequence of time-stamped inspection data recording the damage growth dynamics is premised to this end. Identification of a structural defect in the most recent inspection data set triggers an exhaustive search into the images collected during the previous inspections looking for correspondences based on spatial proximity. This is followed by a view synthesis from multiple candidate images resulting in a single reconstruction for each inspection round. Cracks on concrete surface are used as a case study to demonstrate the feasibility of this approach. Once the chronology is established, the damage severity is quantified at various levels of time scale documenting its progression through time. The proposed scheme enables the prediction of damage severity at a future point in time providing a scope for preemptive measures against imminent structural failure. On the whole, it is believed that the present study will immensely benefit the structural inspectors by introducing the time dimension into the autonomous condition assessment pipeline.

• Journal of the Optical Society of Korea
• /
• v.18 no.3
• /
• pp.217-224
• /
• 2014

A major advantage of wavefront reconstruction based on a series of diffracted intensity images using only single-beam illumination is the simplicity of setup. Here we propose a fast-converging algorithm for wavefront calculation using single-beam illumination. The captured intensity images are resampled to a series of intensity images, ranging from highest to lowest resampling; each resampled image has half the number of pixels as the previous one. Phase calculation at a lower resolution is used as the initial solution phase at a higher resolution. This corresponds to separately calculating the phase for the lower- and higher-frequency components. Iterations on the low-frequency components do not need to be performed on the higher-frequency components, thus making the convergence of the phase retrieval faster than with the conventional method. The principle is verified by both simulation and optical experiments.

• Journal of radiological science and technology
• /
• v.39 no.3
• /
• pp.345-352
• /
• 2016

The purpose of this study was to needed basis of measure MRI CAD development for signal to noise ratio (SNR) by pulse sequence analysis from region of interest (ROI) in brain magnetic resonance imaging (MRI) contrast. We examined images of brain MRI contrast enhancement of 117 patients, from January 2005 to December 2015 in a University-affiliated hospital, Seoul, Korea. Diagnosed as one of two brain diseases such as meningioma and cysts SNR for each patient's image of brain MRI were calculated by using Image J. Differences of SNR among two brain diseases were tested by SPSS Statistics21 ANOVA test for there was statistical significance (p < 0.05). We have analysis socio-demographical variables, SNR according to sequence disease, 95% confidence according to SNR of sequence and difference in a mean of SNR. Meningioma results, with the quality of distributions in the order of T1CE, T2 and T1, FLAIR. Cysts results, with the quality of distributions in the order of T2 and T1, T1CE and FLAIR. SNR of MRI sequences of the brain would be useful to classify disease. Therefore, this study will contribute to evaluate brain diseases, and be a fundamental to enhancing the accuracy of CAD development.

• Journal of Biomedical Engineering Research
• /
• v.18 no.3
• /
• pp.243-251
• /
• 1997

In this paper, we have reported two interesting flow effects arising in the TRFGE sequence using water flow phantom. First, we have shown that the TRFGE sequence is indeed not affected by "in-flow" effect from the unsaturated spins flowing into the imaging slice. Second, the enhancement of "in-plane flow" signal in the readout gradient direction was observed when the TRFGE sequence was used without flow compensation. These two results have many interesting applications in MR imaging other than fMRI. Results obtained were also compared with the results obtained by the conventional gradient echo(CGE) imaging. Experiments were performed at 4.7T MRI/S animal system (Biospec, BRUKER, Switzerland). A cylindrical phantom was made using acryl and a vinyl tube was inserted at the center(Fig. 1). The whole cylinder was filled with water doped with $MnCl_2$ and the center tube was filled with saline which flows in parallel to the main magnetic field along the tube. Tailored RF pulse was designed to have quadratic ($z^2$) phase distribution in slice direction(z). Imaging parameters were TR/TE = 55~85/10msec, flip angle = $30^{\circ}$, slice thickness = 2mm, matrix size = 256${\times}$256, and FOV= 10cm. In-flow effect : Axial images were obtained with and without flow using the CGE and TRFGE sequences, respectively. The flow direction was perpendicular to the image slice. In-plane flow : Sagittal images were obtained with and without flow using the TRGE sequence. The readout gradient was applied in parallel to the flow direction. We have observed that the "in-flow" effect did not affect the TRFGE image, while "in-plane flow" running along the readout gradient direction enhanced the signal in the TRFGE sequence when flow compensation gradient scheme was not used.

• Journal of Korean Neurosurgical Society
• /
• v.63 no.3
• /
• pp.386-396
• /
• 2020

Objective : To generate synthetic spine magnetic resonance (MR) images from spine computed tomography (CT) using generative adversarial networks (GANs), as well as to determine the similarities between synthesized and real MR images. Methods : GANs were trained to transform spine CT image slices into spine magnetic resonance T2 weighted (MRT2) axial image slices by combining adversarial loss and voxel-wise loss. Experiments were performed using 280 pairs of lumbar spine CT scans and MRT2 images. The MRT2 images were then synthesized from 15 other spine CT scans. To evaluate whether the synthetic MR images were realistic, two radiologists, two spine surgeons, and two residents blindly classified the real and synthetic MRT2 images. Two experienced radiologists then evaluated the similarities between subdivisions of the real and synthetic MRT2 images. Quantitative analysis of the synthetic MRT2 images was performed using the mean absolute error (MAE) and peak signal-to-noise ratio (PSNR). Results : The mean overall similarity of the synthetic MRT2 images evaluated by radiologists was 80.2%. In the blind classification of the real MRT2 images, the failure rate ranged from 0% to 40%. The MAE value of each image ranged from 13.75 to 34.24 pixels (mean, 21.19 pixels), and the PSNR of each image ranged from 61.96 to 68.16 dB (mean, 64.92 dB). Conclusion : This was the first study to apply GANs to synthesize spine MR images from CT images. Despite the small dataset of 280 pairs, the synthetic MR images were relatively well implemented. Synthesis of medical images using GANs is a new paradigm of artificial intelligence application in medical imaging. We expect that synthesis of MR images from spine CT images using GANs will improve the diagnostic usefulness of CT. To better inform the clinical applications of this technique, further studies are needed involving a large dataset, a variety of pathologies, and other MR sequence of the lumbar spine.

• Proceedings of the Korean Society of Medical Physics Conference
• /
• 2002.09a
• /
• pp.341-344
• /
• 2002

Neurodegenerative disorders, like Alzheimer's disease, are often accompanied by reduced brain perfusion (cerebral blood flow). Using the intrinsic magnetic properties of water, arterial spin labeling magnetic resonance imaging (ASLMRI) can map brain perfusion without injection of radioactive tracers or contrast agents. However, accuracy in measuring perfusion with ASL-MRI can be limited because of contributions to the signal from stationary spins and because of signal modulations due to transient magnetic field effects. The goal was to optimize ASL-MRI for perfusion measurements in the aging human brain, including brains with Alzheimer's disease. A new ASL-MRI sequence was designed and evaluated on phantom and humans. Image texture analysis was performed to test quantitatively improvements. Compared to other ASL-MRI methods, the newly designed sequence provided improved signal to noise ratio improved signal uniformity across slices, and thus, increased measurement reliability. This new ASL-MRI sequence should therefore provide improved measurements of regional changes of brain perfusion in normal aging and neurodegenerative disorders.