Electrical Engineering

Mobile stations in the cell boundary have poor spectral efficiency due to the path loss and the interference from adjacent cells. Therefore it becomes an important issue to satisfy QoS requirements of each MS at the cell boundary. The deployment of fixed relay stations has been actively considered to solve this spectral efficiency problem at the cell boundary. In this paper, we consider some radio resource management policies that are path selection rule, frequency reuse pattern matching, and frame transmission pattern matching among cells. We evaluate performances of using relays under these policies by varying parameter values such as the relay station's position and frequency reuse factor. Through Monte Carlo simulations and mathematical analysis, we also suggest some optimal values of parameters and policies.

This paper considers some scheduler structures that are executable in the environments of multiple traffic classes and multiple frequency channels. To design a scheduler structure for multiple traffic classes, we first propose a scheduler selection rule that uses the priority of traffic class and the urgency level of each packet. Then we relax the barrier of traffic class priority when a high priority packet has some room in waiting time. This gives us a chance to exploit multiuser diversity, thereby giving more flexibility in scheduling. Our considered scheduler can achieve higher throughput compared to the simple extension of conventional modified largest weighted delay first (MLWDF) scheduler while maintaining the delay performance for QoS class traffic. We also design a scheduler structure for multiple frequency channels that chooses a good channel for each user whenever possible to exploit frequency diversity. The simulation results show that our proposed scheduler increases the total system throughput by up to 50% without degrading the delay performance. Our schedulers are suited to be deployed for OFDMA systems like IEEE 802.16 systems that have plenty of frequency channels and use the adaptive modulation and coding (AMC) scheme.

The dark channel prior is generally a powerful prior for single image dehazing, but it is invalid when there are objects which have similar colors to the atmospheric light. It is difficult to find the accurate transmission factor in this case, and thus the results show some color distortion. This paper proposes a new algorithm for the estimation of correct transmission map regardless of object color. The proposed method defines a reliability map that depicts how much the objects or areas meet the dark channel prior assumption, and then estimate the transmission map using the reliable pixels only. The transmission factors at the unreliable pixels are estimated by linear fitting using the reliable neighboring pixels. Experiments show that the dehazed images by the proposed method have less color distortion than the results by the conventional algorithm.

Under the popular Bayesian approach, a stereo problem can be formulated by defining likelihood and prior. Likelihoods are often associated with unary terms and priors are defined by pair-wise or higher order cliques in Markov random field (MRF). In this paper, we propose to use high order likelihood model in stereo. Numerous conventional patch based matching methods such as normalized cross correlation, Laplacian of Gaussian, or census filters are designed under the naive assumption that all the pixels of a patch have the same disparities. However, patch-wise cost can be formulated as higher order cliques for MRF so that the matching cost is a function of image patch's disparities. A patch obtained from the projected image by a disparity map should provide a better match without the blurring effect around disparity discontinuities. Among patch-wise high order matching costs, the census filter approach can be easily reduced to pair-wise cliques. The experimental results on census filter-based high order likelihood demonstrate the advantages of high order likelihood over independent identically distributed unary model.

When a human matches two images, the viewer has a natural tendency to view the wide area around the target pixel to obtain clues of right correspondence. However, designing a matching cost function that works on a large window in the same way is difficult. The cost function is typically not intelligent enough to discard the information irrelevant to the target pixel, resulting in undesirable artifacts. In this paper, we propose a novel convolutional neural network (CNN) module to learn a stereo matching cost with a large-sized window. Unlike conventional pooling layers with strides, the proposed per-pixel pyramid-pooling layer can cover a large area without a loss of resolution and detail. Therefore, the learned matching cost function can successfully utilize the information from a large area without introducing the fattening effect. The proposed method is robust despite the presence of weak textures, depth discontinuity, illumination, and exposure difference. The proposed method achieves near-peak performance on the Middlebury benchmark.

The conventional methods for estimating camera poses and scene structures from severely blurry or low resolution images often result in failure. The off-the-shelf deblurring or super-resolution methods may show visually pleasing results. However, applying each technique independently before matching is generally unprofitable because this naïve series of procedures ignores the consistency between images. In this paper, we propose a pioneering unified framework that solves four problems simultaneously, namely, dense depth reconstruction, camera pose estimation, superresolution, and deblurring. By reflecting a physical imaging process, we formulate a cost minimization problem and solve it using an alternating optimization technique. The experimental results on both synthetic and real videos show high-quality depth maps derived from severely degraded images that contrast the failures of naïve multi-view stereo methods. Our proposed method also produces outstanding deblurred and super-resolved images unlike the independent application or combination of conventional video deblurring, super-resolution methods.

Removing camera motion blur from a single light field is a challenging task since it is highly ill-posed inverse problem. The problem becomes even worse when blur kernel varies spatially due to scene depth variation and high-order camera motion. In this paper, we propose a novel algorithm to estimate all blur model variables jointly, including latent sub-aperture image, camera motion, and scene depth from the blurred 4D light field. Exploiting multi-view nature of a light field relieves the inverse property of the optimization by utilizing strong depth cues and multi-view blur observation. The proposed joint estimation achieves high quality light field deblurring and depth estimation simultaneously under arbitrary 6-DOF camera motion and unconstrained scene depth. Intensive experiment on real and synthetic blurred light field confirms that the proposed algorithm outperforms the state-of-the-art light field deblurring and depth estimation methods.

Automated breast ultrasound (ABUS) is being rapidly utilized for screening and diagnosing breast cancer. Breast masses, including cancers shown in ABUS scans, often appear as irregular hypoechoic areas that are hard to distinguish from background shadings. We propose a novel branch network architecture incorporating segmentation information of masses in the training process. By providing the spatial attention effect, the branch network boosts the performance of existing neural network classifiers, helping to learn meaningful features around the mass. For the segmentation information, we leverage the existing radiology reports without additional labeling efforts. The reports should include the characteristics of breast masses, such as shape and orientation, and a template mask can be created in a rule-based manner. Experimental results show that the proposed branch network with a template mask significantly improves the performance of existing classifiers.

Custom inspection using X-ray imaging is a very promising application of modern pattern recognition technology. However, the lack of data or renewal of tariff items makes the application of such technology difficult. In this paper, we present a data augmentation technique based on a new image-to-image translation method to deal with these difficulties. Unlike the conventional methods that convert a semantic label image into a realistic image, the proposed method takes a texture map with a special modification as an additional input of a generative adversarial network to reproduce domain-specific characteristics, such as background clutter or sensor-specific noise patterns. The proposed method was validated by applying it to backscatter X-ray (BSX) vehicle data augmentation. The Fréchet inception distance (FID) of the result indicates the visual quality of the translated image was significantly improved from the baseline when the texture parameters were used. Additionally, in terms o...

Automated breast ultrasound (ABUS) is being rapidly utilized for screening and diagnosing breast cancer. Breast masses, including cancers shown in ABUS scans, often appear as irregular hypoechoic areas that are hard to distinguish from background shadings. We propose a novel branch network architecture incorporating segmentation information of masses in the training process. By providing the spatial attention effect, the branch network boosts the performance of existing neural network classifiers, helping to learn meaningful features around the mass. For the segmentation information, we leverage the existing radiology reports without additional labeling efforts. The reports should include the characteristics of breast masses, such as shape and orientation, and a template mask can be created in a rule-based manner. Experimental results show that the proposed branch network with a template mask significantly improves the performance of existing classifiers.

Automated breast ultrasound (ABUS) is being rapidly utilized for screening and diagnosing breast cancer. Breast masses, including cancers shown in ABUS scans, often appear as irregular hypoechoic areas that are hard to distinguish from background shadings. We propose a novel branch network architecture incorporating segmentation information of masses in the training process. The branch network is integrated into neural network, providing the spatial attention effect. The branch network boosts the performance of existing classifiers, helping to learn meaningful features around the target breast mass. For the segmentation information, we leverage the existing radiology reports without additional labeling efforts. The reports, which is generated in medical image reading process, should include the characteristics of breast masses, such as shape and orientation, and a template mask can be created in a rule-based manner. Experimental results show that the proposed branch network with a t...

Under the popular Bayesian approach, a stereo problem can be formulated by defining likelihood and prior. Likelihoods are often associated with unary terms and priors are defined by pair-wise or higher order cliques in Markov random field (MRF). In this paper, we propose to use high order likelihood model in stereo. Numerous conventional patch based matching methods such as normalized cross correlation, Laplacian of Gaussian, or census filters are designed under the naive assumption that all the pixels of a patch have the same disparities. However, patch-wise cost can be formulated as higher order cliques for MRF so that the matching cost is a function of image patch's disparities. A patch obtained from the projected image by a disparity map should provide a better match without the blurring effect around disparity discontinuities. Among patch-wise high order matching costs, the census filter approach can be easily reduced to pair-wise cliques. The experimental results on census filter-based high order likelihood demonstrate the advantages of high order likelihood over independent identically distributed unary model.

The conventional methods for estimating camera poses and scene structures from severely blurry or low resolution images often result in failure. The off-the-shelf deblurring or super-resolution methods may show visually pleasing results. However, applying each technique independently before matching is generally unprofitable because this naïve series of procedures ignores the consistency between images. In this paper, we propose a pioneering unified framework that solves four problems simultaneously, namely, dense depth reconstruction, camera pose estimation, superresolution, and deblurring. By reflecting a physical imaging process, we formulate a cost minimization problem and solve it using an alternating optimization technique. The experimental results on both synthetic and real videos show high-quality depth maps derived from severely degraded images that contrast the failures of naïve multi-view stereo methods. Our proposed method also produces outstanding deblurred and super-resolved images unlike the independent application or combination of conventional video deblurring, super-resolution methods.

Removing camera motion blur from a single light field is a challenging task since it is highly ill-posed inverse problem. The problem becomes even worse when blur kernel varies spatially due to scene depth variation and high-order camera motion. In this paper, we propose a novel algorithm to estimate all blur model variables jointly, including latent sub-aperture image, camera motion, and scene depth from the blurred 4D light field. Exploiting multi-view nature of a light field relieves the inverse property of the optimization by utilizing strong depth cues and multi-view blur observation. The proposed joint estimation achieves high quality light field deblurring and depth estimation simultaneously under arbitrary 6-DOF camera motion and unconstrained scene depth. Intensive experiment on real and synthetic blurred light field confirms that the proposed algorithm outperforms the state-of-the-art light field deblurring and depth estimation methods.

We developed a method to compact the glass sheets of a flat-panel displays that use metal-induced laterally crystallized (MILC) polycrystalline-silicon (poly-Si) thin-film transistors (TFTs), and the effects of thermal stress on the fabricated devices were compared against those of a bare-glass device. The glass substrate was exposed to a temperature of 650 • C for 40 h in order to suppress the glass shrinkage to 0.01 ppm, which suitable for a MILC poly-Si TFT process. The compressive strain that originates from glass shrinkage generally increases the size of the micro-cracks and the vacancies, and as a result, most of the electrical parameters of a bare glass device (such as the on-current, off-current, field-effect mobility, subthreshold slope, and threshold voltage) had a higher level of degradation than those of the device with the compacted glass. The increase in the on-current and the field-effect hole mobility under the compressive strain for poly-Si TFTs showed a similar behavior to that of single-crystalline-silicon (c-Si) TFTs under compressive strain. However, the increase in the off-current was the converse of that of strained c-Si TFT.

Xmas is an extensible main-memory storage system for high-performance embedded database applications. Xmas not only provides the core functionality of DBMS, such as data persistence, crash recovery, and concurrency control, but also pursues an extensible architecture to meet the requirements from various application areas. One crucial aspect of such extensibility is that an application developer can compose application-specific, high-level operations with a basic set of operations provided by the system. Called composite actions in Xmas, these operations are processed by a customized Xmas server with minimum interaction with application processes, thus improving the overall performance. This paper first presents the architecture and functionality of Xmas, and then demonstrates a simulation of mobile communication service.

This paper presents the architecture of Xmas, an extensible mainmemory storage system for high-performance and real-time database applications. Xmas not only provides the core functionality of a DBMS, such as data persistence, concurrency control, and crash recovery, but aIso has an extensible architecture to deal with requirements from various applications. One crucial aspect of such extensibility is that an application developer can compose applicationspecific, high-level operations with a basic set of operations provided by the system. CalIed composite actions in Xmas, these operations are processed by a customized Xmar server with minimum interaction with application processes, thus increasing the overall performance. 1 Introduction Many applications such as mobile communication systems, factory automation, and distributed virtual reality networks, require high-performance and real-time access to database[KGM95]. Disk-based DBMSs, however, are inadequate for these types of applications because of high disk access overhead for data retiewl and update. Mainmemory DBMSs, which became feasible with the increasing availability of Iarge and relatively cheap memory, are expected to support the above applications better than diskbased ones[GMS92]. Xmas is an eXtensibIe Main-memory Storage system, which has been under recent development at Seoul. National University for high-performance and real-time database applications[CPL+96], [CPL+95]. Xmas provides the core functionality of a DBMS, data persistence, concurrency control, and crash recovery. Furthermore, it pursues an extensible architecture to deal with requirements from various applications. We argue that the extensibility of a main-memory I'cnnision to make digit:M~;~rd topics nl'nll or part of this mat&-d Ibr pzrsond or c'J;t.woon~ ww is gran~cd wilhaut Ike provided that the cupim are not made or distributed ibr prolil w comnxrciat nd~~~tngc. the atpyright novices 111r: tide ol'ths publicatiw mrd its da~c ;yqw~r. and mtice is givm thl mpyright is iy pcmissimt ut'thr'/ZCRI. Inc. 'I'0 copy olhwavisc~ to rqxthlidr. to post an s~vas or to raktribttk 10 lists. rquir* spxilic pennissiot~ and/or tir, CIKM 97 Lnsr4qm !vclwc/cz Iis.