• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2019.11a
• /
• pp.252-255
• /
• 2019

본 논문에서 Equirectangular projection(ERP) 영상을 행동 인식 모델에 입력하기전 제안하는 전처리를 통하여 성능을 향상시키는 것을 보인다. ERP 영상의 특성상 행동 인식을 하는데 불필요한 영역이 일반적인 2D 카메라로 촬영한 영상보다 많다. 또한 행동 인식은 사람이 Object of Interest(OOI)이다. 따라서 객체 인식모델로 인간 객체를 인식한 후 Region of Interest(ROI)를 추출하여 불필요한 영역을 없애고, 왜곡 또한 줄어든다. 본 논문에서 제안하는 기법으로 전처리 후 CNN-LSTM 모델로 성능을 테스트했다. 제안하는 방법으로 전처리를 한 데이터와 하지 않은 데이터로 행동 인식을 한 정확도로 비교하였으며 제안하는 기법으로 전처리 한 데이터로 행동 인식을 한 경우 데이터의 특성에 따라 다르지만, 최대 61%까지 성능향상을 보였다.

• Bulletin of the Korean Chemical Society
• /
• v.22 no.5
• /
• pp.488-492
• /
• 2001

As gas chromatography/infrared spectrometry (GC/IR) becomes routinely avaliable, methods must be developed to deal with the large amount of data produced. We demonstrate computer methods that quickly search through a large data file, locating thos e spectra that display a spectral feature of interest. Based on a modified library search routine, these selective data reduction methods retrieve all or nearly all of the compounds of interest, while rejecting the vast majority of unrelated compounds. To overcome the shifting problem of IR spectra, a search method of moving the average pattern was designed. In this moving pattern search, the average pattern of a particular functional group was not held stationary, but was allowed to be moved a little bit right and left.

• Journal of the Korea Safety Management & Science
• /
• v.5 no.4
• /
• pp.267-278
• /
• 2003

Data Envelopment Analysis-Assurance Region(DEA-AR) model is used in this paper to investigate the efficiency and performance potential of Korean banks as they engage in activities that incur interest and non-interest expenses and produce income. DEA provides a measure of each bank's relation to the best-practice frontier for its competitors. This can provide a better quality-benchmark than using industry averages or a particular peer bank branches as the benchmark. The banks are classified into efficient and inefficient sets. Multiplier values for AR-inefficient banks with unique slacks indicate the potential for management to improve the bank's performance relative to its peers. DEA-AR that provide economically reasonable bounds for the multipliers lead to profitability potential, as distinct from efficiency, results.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.28 no.6C
• /
• pp.634-647
• /
• 2003

In MPEG-4 FGS (fine granular scalability) video, SE (selective enhancement) function is adopted to enhance the subject quality of the region of interest (ROI). However, it has the problem of excessive bit-rate increase in the enhancement layer. We present a new rectangular region-based SE (RSE) algorithm to significantly reduce the overhead bits resulting from the standard SE. The proposed RSE is based on two new algorithms. The first is to apply the SE function to a rectangular region. By doing so, we can reduce the required bits for describing the selectively enhanced region. The second is to use constrained bit-plane scanning (CBS) to encode bit-planes of the enhancement layer. By using CBS, we can efficiently encode the ALL-ZERO symbols that are generated by applying the SE. It Is shown by simulation that the proposed RSE can provide a good visual quality for the selected rectangular region with significantly reduced overhead bits.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.28 no.4
• /
• pp.421-430
• /
• 2010

Image segmentation technique is becoming increasingly important in the field of remote sensing image analysis in areas such as object oriented image classification to extract object regions of interest within images. This paper presents a new method for image segmentation in High Resolution Remote Sensing Image based on Improved Seeded Region Growing (ISRG) and Region merging. Firstly, multi-spectral edge detection was done using an entropy operator in pan-sharpened QuickBird imagery. Then, the initial seeds were automatically selected from the obtained multi-spectral edge map. After automatic selection of significant seeds, an initial segmentation was achieved by applying ISRG to consider spectral and edge information. Finally the region merging process, integrating region texture and spectral information, was carried out to get the final segmentation result. The accuracy assesment was done using the unsupervised objective evaluation method for evaluating the effectiveness of the proposed method. Experimental results demonstrated that the proposed method has good potential for application in the segmentation of high resolution satellite images.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.10 no.6
• /
• pp.1153-1158
• /
• 2006

The classification of the background and cell areas is very important research area because of the ambiguous boundary. In this paper, the region of cell is extracted from an image of uterine cervical cytodiagnosis using the region growing method that increases the region of interest based on similarity between pixels. Segmented image from background and cell areas is binarized using a threshold value. And then 8-directional tracking algorithm for contour lines is applied to extract the cell area. First, the extracted nucleus is transformed to RGB color that is the original image. Second, the K-means clustering algorithm is employed to classify RGB pixels to the R, G, and B channels, respectively. Third, the Hue information of nucleus is extracted from the HSI models that is the transformation of the clustering values in R, G, and B channels. The backpropagation algorithm is employed to classify and identify the normal or abnormal nucleus.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.53 no.11
• /
• pp.41-47
• /
• 2016

High efficiency video coding (HEVC) employs a coding tree unit (CTU) to improve the coding efficiency. A CTU consists of coding units (CU), prediction units (PU), and transform units (TU). All possible block partitions should be performed on each depth level to obtain the best combination of CUs, PUs, and TUs. To reduce the complexity of block partitioning process, this paper proposes the PU mode skip algorithm with region of interest (RoI) selection using motion vector. In addition, this paper presents the CU depth level skip algorithm using the co-located block information in the previously encoded frames. First, the RoI selection algorithm distinguishes between dynamic CTUs and static CTUs and then, asymmetric motion partitioning (AMP) blocks are skipped in the static CTUs. Second, the depth level skip algorithm predicts the most probable target depth level from average depth in one CTU. The experimental results show that the proposed fast CU decision algorithm can reduce the total encoding time up to 44.8% compared to the HEVC test model (HM) 14.0 reference software encoder. Moreover, the proposed algorithm shows only 2.5% Bjontegaard delta bit rate (BDBR) loss.

• Journal of Broadcast Engineering
• /
• v.11 no.3 s.32
• /
• pp.336-348
• /
• 2006

The recent advances in multimedia signal coding and transmission technologies allow lots of users to watch videos on small LCD displays. In this paper, we briefly describe an intelligent display technique to provide small-display-viewers with comfortable experiences, and study the minimum image size tolerated and utility of displaying region of interest (ROI) only when needed. The study, with 111 participants, examines minimum image size to ensure viewers pleasant viewing experiences, and evaluates the degree of satisfaction when they are viewed with region of interest (ROI) only. The experimental results show that the ROI display enhances the viewers' satisfaction when the image size becomes less than $320{\times}240$, and thus it is useful to provide the intelligent display, if necessary, which can extract and display ROI only.

• Journal of the Korea Computer Graphics Society
• /
• v.23 no.3
• /
• pp.65-75
• /
• 2017

As a 360-degree image carries information of all directions, it often has too much information. Moreover, in order to investigate a 360-degree image on a 2D display, a user has to either click and drag the image with a mouse, or project it to a 2D panorama image, which inevitably introduces severe distortions. In consequence, investigating a 360-degree image and finding an object of interest in such a 360-degree image could be a tedious task. To resolve this issue, this paper proposes a method to find a region of interest and produces a 2D naturally looking image from a given 360-degree image that best matches a description given by a user in a natural language sentence. Our method also considers photo composition so that the resulting image is aesthetically pleasing. Our method first converts a 360-degree image to a 2D cubemap. As objects in a 360-degree image may appear distorted or split into multiple pieces in a typical cubemap, leading to failure of detection of such objects, we introduce a modified cubemap. Then our method applies a Long Short Term Memory (LSTM) network based object detection method to find a region of interest with a given natural language sentence. Finally, our method produces an image that contains the detected region, and also has aesthetically pleasing composition.

• The KIPS Transactions:PartD
• /
• v.10D no.7
• /
• pp.1171-1176
• /
• 2003

In this paper, we propose an effective bar code detection algorithm using the feature analysis and the labeling. After computing the direction of pixels using four line operators, we obtain the histogram about the direction of pixels by a block unit. We calculate the difference between the maximum value and the minimum value of the histogram and consider the block that have the largest difference value as the block of the bar code region. We get the line passing by the bar code region with the selected block but detect blocks of interest to get the more accurate line. The largest difference value is used to decide the threshold value to obtain the binary image. After obtaining a binary image, we do the labeling about the binary image. Therefore, we find blocks of interest in the bar code region. We calculate the gradient and the center of the bar code with blocks of interest, and then get the line passing by the bar code and detect the bar code. As we obtain the gray level of the line passing by the bar code, we grasp the information of the bar code.