Hong Jeong

• PhD
• Professor at Inha University in Tashkent

Green bond issuance is a highly technical and challenging procedure involving many parties with conflicting goals. In this chapter, a framework for digitalization of green bonds is provided. We propose a blockchain-enabled green bond-issuing platform that will provide an adaptive and efficient system that lowers intermediary costs by reducing infor...

We propose a fast image-dehazing method that uses a pixelwise dark channel prior (PDCP). We argue that the neighbor minimum filter in the dark channel prior (DCP) is not crucial in image dehazing. We prove this assertion in experiments that compare dehazed images and histograms of dark channel images obtained using PDCP and DCP. To refine the trans...

Detecting moving objects in a dynamic scene is a difficult task in computer vision. We propose a moving object detection algorithm for advanced driver assistance systems that uses only images from a monocular camera. To distinguish moving objects from standing objects when the camera is moving, we used an epipolar line constraint and an optical flo...

This study investigates the directed acyclic subgraph (DAS) algorithm, which is used to solve discrete labeling problems much more rapidly than other Markov-random-field-based inference methods but at a competitive accuracy. However, the mechanism by which the DAS algorithm simultaneously achieves competitive accuracy and fast execution speed, has...

We proposed an algorithm that uses an energy model with smoothness assumption to identify a moving object by using optical flow, and uses a particle filter with a proposed observation and dynamic model to track the object. The algorithm is based on the assumption that the dominant motion is background flow and that foreground flow is separated from...

We present an improved non-learning-based confidence measure based on tree agreement for the stereo matching problem. To use confidence information for accurate matching, we propose an improved method to assemble a cost aggregation table with confidence measure. The proposed confidence measure and cost aggregation method were evaluated using the KI...

In advanced driver assistance systems, the stereo matching algorithm is the key resource to obtain depth information of outdoor scenes. Semi-Global Matching (SGM) is currently the most efficient stereo matching algorithm for outdoor environments. However, because the number of pixels is large, SGM uses only a subset of them when estimating the disp...

We present a method that can detect a circular traffic signs and Light Emitting Diode (LED) traffic signs by combining fast radial symmetric transform and a proposed symmetric shape detection method. The proposed method can choose the appropriate size and sign type by computing the confidence of given image to each pre-defined signs, In a real road...

The stereo matching algorithms usually do not satisfy the performance of both accuracy and complexity. The semi-global matching (SGM) is the most efficient method among the stereo matching algorithms. However, it still does not utilize the whole image information to estimate the disparity values. In this paper, we proposed a new concept of the cost...

This book provides comprehensive coverage of 3D vision systems, from vision models and state-of-the-art-algorithms to their hardware architectures for implementation on DSPs, FPGA and ASIC chips and GPUs. It aims to fill the gaps between computer vision algorithms and real-time digital circuit implementations, especially with Verilog HDL design. Th...

This chapter introduces the concept underlying the DP algorithm and architecture, and explains the computational structures of various DP algorithms.

This chapter deals with Verilog design of serial architectures for the DP. The design is aimed at executing stereo matching with the serial vision simulator.

This chapter introduces the systolic array in Verilog HDL. It explains in detail how to design the control module and the systolic array, connected by local neighborhood connections.

This chapter deals with the BP design for stereo matching. It explains in detail the design methods with Verilog HDL.

This chapter deals with motion estimation and fusion of vision modules. Instead of giving an extensive survey, it focuses on the motion principles and the continuity concept that unify the various constraints in motion estimation. It also deals with the fusion of vision modules, directly with intermediate variables bypassing the 3D variables, which...

This chapter introduces the basics of Verilog HDL and coding examples for communication and control modules. These modules are general building blocks for designing vision architectures.

This chapter introduces the concept underlying the relaxation algorithm and architecture. In addition to the Gauss–Seidel and Jacobi algorithms, it introduces other types of architectures; specifically, extensions to the Gauss–Seidel–Jacobi architecture.

In this paper, a new Advanced Driver Assistant System (ADAS) system for LED traffic sign detection algorithm using rectangle shape based on a windowed hough transform and feature based optimization was presented. We used two character to detect the LED traffic sign. One is LED traffic sign have rectangle shape, another is intensity feature of LED t...

In this paper, we present a novel matching framework for stereo video sequence based on disparity prediction by stereo-motion fusion. Fusion-based methods tackles the problem by integrating different depth modalities in cooperative way, so as to resolve matching ambiguities from each modality and acquire more accurate disparity map. Unfortunately,...

In synthesizing intermediate views from two stereo images with the help of stereo matching, we suggest an efficient method for removing occlusion. The proposed method can deal with the problems: noise correction from stereo matching, occlu-sion removal for high fidelity rendering, and parallel computing for real-time synthesis. The proposed algorit...

We suggest a relationship, called stereo-motion equation, between stereo disparity and optical flow, and a recursive filter, as an efficient algorithm to estimate the two quantities. We show that close spatial and temporal relationships exist between the two quantities. The importance of this discovery is that the vision quantities can be computed...

We suggest an efficient real-time method for removing occlusions when synthesizing intermediate views from two stereo images by stereo matching algorithm. The proposed method can carry out the correction of noise resulting from stereo matching, removal of occlusions to achieve high fidelity rendering, and parallel computing to achieve real-time syn...

In this paper, we present an Augmented Reality (AR) system using Quick Responsible (QR) code as the marker for Android Smartphone. QR code has many advantages to be a marker. It can encode relatively larger amount of marker information in an easy and standard way, also it has the capability of error correction. Basically the system detects the mark...

Up to these days, there are huge amount of stereo matching algorithms and its improved versions. Not sufficient algorithms are proposed to be used in real systems like object recognition not considering time computing and complexity of the algorithm, but only for improving Pixel Error Rate (PER) in disparity image. We used some image processing met...

To make real-time stereo matching system is the last goal in stereo vision area. we tried to make this system that shows sufficient to be used for many application but not perfect in performance because of the limitation in hardware design. We used trellis algorithm which is advanced version of the Dynamic Programming algorithm, and we implemented...

This paper present a method to measure Line-edge Roughness in Scanning Electron Microscope images with high accuracy based on the digital image processing. As lithography source will be changed from ArF to Extreme UltraViolet (EUV), this trend increases the importance and difficulty of measuring LER. To measure LER accurately, we use two method; im...

Currently, many different defect detection algorithms have developed. Most finds defect using optical method, and some using image processing uses a reference image. There are several problems using the reference image. The scale, illumination or orientation difference between the reference and the target image make it hard or even impossible to fi...

One of the problems in stereo matching algorithm is that it is impossible to estimate the most left positioned pixels with left reference system or the most right positioned pixels with right reference system because of the characteristic of stereo vision. That means in stereo matching algorithm, it cannot estimate disparity in occlusion parts. We...

In synthesizing intermediate views from two stereo images with the help of stereo matching, we suggest an efficient method for removing occlusion. The proposed method can deal with the problems: noise correction from stereo matching using photometric method, occlusion removal for high fidelity rendering using disparity consistency, and parallel com...

Stereo vision area is widely studied for long times to make accurate and to work in real-time. DP algorithm is chosen among many stereo matching algorithms to make real-time stereo matching system. And some tricky image processing methods are used to suppress streak noise, flaw of DP. Whole procedure consists of three parts, calibration and rectifi...

As a convergence method for stereo matching and motion estimation, this paper presents an equation , called Disparity-Optical flow Equation, that relates disparity with optical flow in rectified images. Considering this equation as a constraint, this paper suggests an algorithm, called Simultaneous Disparity and Optical Flow under Epipolar Constrai...

Multilayer perceptrons trained with the backpropagation algorithm are tested in gun fire control system for error correction and are compared to optimal algorithms based on minimum mean square error.1,2 The structure of the proposed neural controller is described and performance results are shown.

There is a growing need in computer vision applications for stereopsis, requiring not only accurate distance but also fast and compact physical implementation. Global energy minimization techniques provide remarkably precise results. But they suffer from huge computational complexity. One of the main challenges is to parallelize the iterative compu...

To extract emotion from speech signals, we must specify the representation and grammatical model, which are still challenging issues. We proposed a new feature called Non- negative Matrix Factorization (NMF) feature. The proposed algorithm has been tested in several different ways by varying NMF and the speech database. We compared the recognition...

We present a novel visual obstacle detection and tracking system based on stereo vision. A robust stereo matcher, an obstacle detector and a tracker module are implemented and tested under actual driving conditions. Implemented system shows reliable range estimation, detection and tracking performance. Our system has showed 82.4 percent correctly d...

Stereo vision is the process of recovering 3D spatial information from a pair of 2D images. It is very difficult problem due to the fact that stereo matching problem tends to produce as large number of plausible solutions. Thus we need to restrict the solution space in some manner. Trellis-based stereo matching algorithm places hard constraints on...

There has been lots of studies about stereo vision area to get more better depth images. However there are not sufficient amount of researches about stereo matching algorithm using pyramid structure. We proposed some methods to get high quality depth image using Gaussian window matching cost, pyramid structure, and median filtering. These methods s...

In this paper, we propose a segmentation algorithm for separating focused foreground region from defocused background region with low depth of field(DOF) image. Low DOF is a one of the most popular photography technique for emphasize the importance object. and highly useful in various application including image enhancement for digital cameras, tar...

In the semiconductor industry, detecting defects in mask patterns is very important, because even a single defect may lead to failure. Thus, controlling and detecting defects are necessary steps in the manufacturing process. Since there are numerous types of mask patterns, a universal automatic method for defect recognition has not been developed....

This paper derives the disparity-blur equation, which describes an explicit relationship between stereo disparity and blur and thereby suggests a method for depth estimation by combining the two quantities. Because disparity and blur are two different aspects of the same 3D point, a direct geometric relationship in the image plane exists between th...

In this paper, we propose a new segmentation algorithm for separate focused foreground region from defocused background region with low depth of field image. This algorithm first computes the blur parameter each of pixel. Then, these parameter values are clustered to blurry regions and sharp regions using belief propagation algorithm. Experimental...

The purpose of Blind Source Separation (BSS) is to obtain separated sources from convolutive mixture inputs. Among the various available BSS methods, Independent Component Analysis (ICA) is one of the representative methods. Its key idea is to repetitively update and calculate the measures. However, dealing with the measures obtained from multi-arr...

We propose a kind of source separating method in audio signal. It uses an orthogonal property in spectral domain, which is an extended concept of temporal orthogonality used in Degenerate Unmixing Estimation Technique (DUET) to individual frequency bands. In addition, we also adopt a gradient-based learning rule using negentropy for perfect separat...

In this paper, we suggest the relationship, called disparity-optical flow equation, between the two vision quantities: stereo disparity and optical flow, and a recursive filter, as an efficient algorithm. It is shown that, in a general setting, there exist close spatial and temporal relationships between the two quantities. The importance of this d...

In this paper, we propose a new algorithm, which is using NMF features and which is different from traditional algorithms using temporal variations of the spectrogram. The proposed algorithm has been tested in several different ways by varying NMF and the speech database. The experimental results show performance of 90% and classification success o...

We propose a source extraction algorithm that provides robust speech recognition of intelligent service robots. This system consists of two stages. First, the target speech from a target direction is extracted from a mixture inputs using a binary time-frequency masking method. This stage can work alone. The next stage re-estimates the range of the...

There are numerous algorithms to detect emotion from speech signals. Among the algorithms, we selected spectral analysis and fused Non-negative Matrix Factorization (NMF) for obvios emotion classification. The algorithm has been tested in several different ways by varying NMF and the speech database. The experimental results show performance of 90%...

In this paper, we present a new and automated technique to extract defects in photomask images. To correctly extract defects, we propose a robust automated method based on second order moment (SOM) between reference and test images and a statistical model based on difference image are used. The statistical model is distribution of the normalized ab...

In this paper, we propose a speech recognition system with a narrow sound channel for intelligent service robots in noisy environments. The narrow sound channel is obtained by using time delay between two array microphone inputs used in the degenerate unmixing estimation technique (DUET). In the proposed system, the voice from a specific direction...

In this study, we investigate the orthogonal properties between independent audio sources in time-frequency do-main and suggest a separation algorithm using the property. The proposed algorithm adaptively uses some existing techniques including Frequency Domain-Independent Component Analysis (FD-ICA) in hybrid manner. The experimental section rep-r...

This paper describes an automated technique to extract defects in photomask images. Conventional approaches to photomask inspection rely on difference based techniques which are susceptible to distortion from image `noise'. We propose a robust method based on high order moment (HOM) between reference and test images. In comparison to difference bas...

There are growing needs in computer vision applications for stereo matching, requiring not only accuracy and robustness but also fast processing speed. Global matching algorithms such as belief propagation (BP) shows remarkably robust results with presence of occlusions, textureless region, image noise. In this paper, a novel approach that combines...

Intermediate view synthesis is an important step for Multi-view 3D displays. In this paper we have proposed an intermediate view synthesis approach, which is based on stereo matching process. This makes that multi-view images can be created by only the two-view images, which are captured by two-view camera(stereo camera).In order to obtain accurate...

This paper describes a new method for the process of extracting photomask defects using a single optical photomask image which is not aligned. The lines of a photomask picture are not parallel with the pixel grid, and so there is always a angle of error. Only one sample image which contains defects was used in our tests. The algorithm is effective...

This paper presents a real-time aging measurement and feasible compensation system for the prolonged lifetime of OLED panel. The proposed system is composed of four parts, a PC with a man-machine interface, a measurement block, an adaptive amplifier block, and a compensation block. We apply a tree algorithm for less complexity and convenience of me...

In this paper we compare two different blind source separation algorithms, Degenerate Unmixing Estimation Technique (DUET) and Frequency-Domain Independent Component Analysis (FDICA). Even though both algorithms are known to be powerful in separating signals, each has superior to another in a specific condition. We investigate advantages and disadv...

Although stereo matching algorithms based on belief propagation (BP) tend to show excellent matching performance, their huge computational complexity has been the major barrier to real-time applications. In this light, we propose a parallel very large scale integration (VLSI) architecture for BP computation, which has only simple integer operations...

In optical flow, it is well known that the global matching methods such as belief propagation (BP) and graph cut provide remarkably robust and precise results in low level vision. Unfortunately, all these methods suffer from huge computational complexity due to iterations over the entire pixels. We have observed that changing the order of iterative...

There are growing needs in computer vision applications for stereopsis, requiring not only accurate distance but also fast and compact physical implementation. Presently, the global matching techniques provides remarkably robust and precise results in stereo matching. Unfortunately, they suffer from huge computational and memory complexity. We have...

As a step towards real-time stereo, we will present a fast and efficient VLSI architecture and implementation of a stereo matching algorithm which has a low error rate. The architecture has the form of linear systolic array using simple processing element(PE)s that are connected with neighboring PEs. Due to this simple full parallel structure, it i...

In this paper we consider the problem of sensing the mixture of sources in a distributed network of sensors. We assume that sensors have low computation power. They quantize the mixtures and send them to the server for separation. We investigate the system performance using two low complexity blind source separation methods, FastICA and JADE. FastI...

Successful speech recognition in noisy environments for intelligent robots depends on the performance of preprocessing elements employed. Even though acoustic signals are often corrupted in the high noise level environment, speech recognition systems such as the widely-used HTK do not deal with signal distortion problems. We propose an architecture...

Toward real-time mean shift, a high-speed and parallel mean shift filter on a 2D regular graph is presented in this paper. For an by image and with iteration times, ´ µ time complexity of sequential computation is re- duced to ´ µ with processors, and ´ µ memory complexity is reduced to ´ µ when is smaller than . As a result, computational speed is...

Toward real-time mean shift, a high-speed and parallel mean shift filter on a 2D regular graph is presented in this paper. For an M by N image and with L iteration times, O(NML) time complexity of sequential computation is reduced to O{NL) with M processors, and O(NM) memory complexity is reduced to O(LM) when L is smaller than N. As a result, comp...

As a step towards real-time stereo on 2D markov random field (MRF), we will present fast belief propagation (FBP) VLSI architecture for stereo matching, which has a parallel, distributed and memory-efficient structure and lowest error rates among the real-time systems. FBP can reduce memory complexities by 17 times smaller than belief propagation (...

As a step towards real-time stereo, we will present a fast and efficient VLSI architecture and implementation of a stereo matching algorithm which has a low error rate. The architecture has the form of linear systolic array using simple processing element (PE)s that are connected with neighboring PEs. Due to this simple full parallel structure, it...

Blind source separation of independent sources from their convolutive mixtures is a problem in many real-world multi-sensor applications. However, the existing BSS architectures are more often than not based upon software and thus not suitable for direct implementation on hardware. The existing software of feedback network algorithm is not suitable...

In this paper, we present an efficient architecture for connected word recognition that can be implemented with field programmable gate array (FPGA). The architecture consists of newly derived two-level dynamic programming (TLDP) that use only bit addition and shift operations. The advantages of this architecture are the spatial efficiency to accom...

This paper describes the process of extracting and classifying defect in optical, photomask images. It describes photomask defect extraction which is the preprocessing of classification. In addition, it describes not only how to classify defect into false, particle on glass, and pattern defect but also how to classify pattern defect into spot, Cr e...

This paper addresses the problem of dual-channel speech enhancement based on noise analysis in real time. In previous algorithms related to the speech intelligibility, the improvement in intelligibility is mostly incorporated in a single-channel environment where the speech and noise signals are mixed together. In this paper, we deal with the dual-...

In this paper, we propose an algorithm which enhances the speech intelligibility using the properties of human auditory system. In previous algorithms related to the speech intelligibility, the improvement in intelligibility has been mostly incorporated in a single-channel environment where the speech and noise signals are mixed together. But the s...

The stereo matching algorithm based on the belief propagation (BP) has the low matching error as the global method, but has the disadvantage of a long processing time. In addition to a low error in the Middlebury image simulation, a new architecture based on BP shows a high-speed parallel VLSI structure of the time complexity O(N), at properly smal...

The stereo matching algorithm based on the belief propagation (BP) has the low matching error as the global method, but has the disadvantage of a long processing time. In addition to a low error of less than 2.6% in the Middlebury image simulation, a new architecture based on BP shows a high-speed parallel VLSI structure of the time complexity O(N)...

The structure analysis of pop and rock songs from audio signals is conducted via similarity matrix processing in this work. The similarity matrix offers pairwise similarity be-tween any two short intervals of fixed length in a song. We use two similarity matrices to show their diverse charac-teristics. The characteristics are explained by musical c...

In this paper, we present an efficient architecture for connected speech recognition that can be efficiently implemented with FPGA. The architecture consists of newly derived two-level dynamic programming(TLDP) that use only bit addition and shift operations. The advantages of this architecture are the spatial efficiency to accommodate more words w...

ABSTRACT The analysis of audio signals of popular and rock songs of the verse- chorus form to reconstruct its original musical structures is investi- gated in this work. We first compute,the similarity degree bet ween any two measures,in a song based on selected features and represent these numbers in a measure-based similarity matrix. Then, we stu...