• 한국멀티미디어학회논문지
• /
• 제24권3호
• /
• pp.373-381
• /
• 2021

In this paper, we propose a classification model based on convolutional neural network(CNN) for predicting 2-year recurrence in non-small cell lung cancer(NSCLC) patients using preoperative chest CT images. Based on the region of interest(ROI) defined as the tumor internal and external area, the input images consist of an intratumoral patch, a peritumoral patch and a peritumoral texture patch focusing on the texture information of the peritumoral patch. Each patch is trained through AlexNet pretrained on ImageNet to explore the usefulness and performance of various patches. Additionally, ensemble learning of network trained with each patch analyzes the performance of different patch combination. Compared with all results, the ensemble model with intratumoral and peritumoral patches achieved the best performance (ACC=98.28%, Sensitivity=100%, NPV=100%).

• 대한의용생체공학회:의공학회지
• /
• 제26권1호
• /
• pp.43-47
• /
• 2005

We have developed a small bore x-ray CT for small animal imaging with a linear x-ray detector array and small-scale slip rings. The linear x-ray detector array consists of 1024 elements of 400□m×400□m with a gadolinium oxysulfide (GOS) scintillator on top of them. To avoid use of expensive large diameter slip rings for projection data transmission from the X­ray detector to the image reconstruction system, we used the wireless LAN technology. The projection data are temporally stored in the data acquisition system residing on the rotating gantry during the scan and they are transmitted to the image reconstruction system after the scan. With the wireless LAN technology, we only needed to use small-scale slip rings to deliver the AC electric power to the X-ray generator and the power supply on the rotating gantry. The performances of the small animal CT system, such as SNR, contrast, and spatial resolution, have been evaluated through experiments using various phantoms. It has been experimentally found that the SNR is almost linearly proportional to the tube current and tube voltage, and the minimum resolvable contrast is less than 30 CT numbers at 40kVp/3.0㎃. The spatial resolution of the small animal CT system has been found to be about 0.9Ip/㎜. Postmortem images of a piglet is also presented.

• 항공우주시스템공학회지
• /
• 제16권5호
• /
• pp.70-77
• /
• 2022

본 논문에서, 데이터베이스 생성 및 관리 시스템은 미소 픽셀 공중 표적 인식을 위해 제안된다. 제안된 시스템은 1)비행 테스트 비디오 프레임에 의한 직접 이미지 추출, 2) 자동 이미지 보관, 3) 이미지 데이터 레이블링 및 메타 데이터 주석, 4) 컬러 채널 변환, 5) HOG/LBP 기반 소화소 대상 증강 이미지 데이터 생성의 다섯가지 주요 기능으로 구성된다. 제안하는 프로그램은 파이썬 기반의 PyQt5와 OpenCV를 이용하여 구성하였고 공중 표적 인식을 위한 이미지 데이터셋은 제안한 시스템을 이용해 생성했으며 비행 실험으로 부터 수집된 영상을 입력영상으로 사용하였다.

• 대한기계학회논문집B
• /
• 제24권2호
• /
• pp.153-160
• /
• 2000

A Particle Image Velocimetry(PIV) algorithm is developed to analyze whole flow field both qualitatively and quantitatively. The practical use of PIV requires the use of fast, reliable, computer-based methods for tracking numerous particles suspended in a flow field. The TSS, NTSS, FFT-Hybrid, which are developed in the area of image compression and coding, are introduced to develop fast vector search algorithm. The numerical solution of the lid-driven cavity flow by the ADI algorithm with the Wachspress Formula is introduced to produce synthetic data for the validation of the tracking algorithms. The algorithms are applied to image data of real flow experiments. The comparisons in CPU time and mean error show, with a small loss of accuracy, CPU time for tracking is reduced considerably.

• 한국의류학회지
• /
• 제30권1호
• /
• pp.137-145
• /
• 2006

This study offers the basis of the image making in categorizing one's personal style and investigation its features and image differences, according to the image making on the subject of Korean female entertainers. The image factors are Activeness, Attractiveness, Extroversions and Intimacy that has occupied by $56.34\%$ out of total. The type of the personal style of Korean female entertainers are classified into five groups and nine small groups that refer to Elegant style(Noble, Sophisticated), Natural style, Tough style, Romantic style(Cute, Delicate), Sexy style(Vivid, Cool, Healthy). The image according to the type of personal style has presented meaningful differences in all composed factors, in particular, the most significant differences are shown in the factors of Attractiveness and Activeness. The axis of the personal positioning far Korean female entertainers are divided into Elegant-Active, Cool-Warm, Weak-Tough. Artificial -Natural.

• 제어로봇시스템학회논문지
• /
• 제13권12호
• /
• pp.1160-1166
• /
• 2007

This paper presents the design and implementation of an embedded image stabilization control system for a mobile phone with a built-in camera. Image stabilization is a family of techniques for reducing image blur resulting from minute camera shake due to hand-held shooting, thereby allowing the use of shutter speeds slower than values normally required to obtain sharp images. A mechanical image stabilizer mechanism developed for a camera mobile phone is introduced and a digital control system as a part of the image stabilization system is designed and implemented on an 8-bit microcontroller with integer arithmetic in C. This paper focuses primarily on issues that need to be taken into consideration for fixed-point implementation of the digital controller. Several experimental results are presented to demonstrate the performance of the implemented image stabilization control system.

• 한국시뮬레이션학회논문지
• /
• 제10권3호
• /
• pp.59-70
• /
• 2001

Typical low-level image processing tasks require thousands of operations per pixel for each input image. Traditional general-purpose computers are not capable of performing such tasks in real time. Yet important features of traditional computers are not exploited by low-level image processing tasks. Since storage requirements are limited to a small number of low-precision integer values per pixel, large hierarchical memory systems are not necessary. The mismatch between the demands of low-level image processing tasks and the characteristics of conventional computers motivates investigation of alternative architectures. The structure of the tasks suggests employing an array of processing elements, one per pixel, sharing instructions issued by a single controller. In this paper we implemented various image processing filtering using the format converter. Also, we realized from conventional gray image process to color image process. This design method is based on realized the large processor-per-pixel array by integrated circuit technology This format converter design has control path implementation efficiently, and can be utilize the high technology without complicated controller hardware.

• 한국조경학회지
• /
• 제32권1호
• /
• pp.47-56
• /
• 2004

Rapid economic development in Korea caused functions of city functions such as concentration of population, deterioration of the quality of living environment and traffic congestion. Korean cities have lost their identity becausr they are merged functionally and physically with neighboring cities, forming one mesa-city. Unified shape and disorganized streets of cities often cause confusion among foreigners and visitors. It is very difficult for them to find their image in strange cities. It is, however, important to correctly analyze the image and meaning of cities for understanding its identity. The purpose of this study is to develop a method to analyze the city image by focusing on some of the main cities in Korea. For this purpose, the adjective questionnaire and multi-dimension scaling (MDS) are applied to the analysis of city image. Image analysis graph by MDS can visually present the general and integrate images. The results of this study are summarized as follows: The important factors for interpretation of city image are historical and industrial character. Seoul, Taegu and Pusan have industrial and complex city images. Kongju has historical city image, while Changwon has a modern image. Chuncheon belongs to a soft and small image. Each city has an alternative solution against a negative image, according to the image analysis graph.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2002년도 하계종합학술대회 논문집(4)
• /
• pp.211-214
• /
• 2002

A novel structuring element for the morphological filter is proposed in order to detect a small target at a long distance. The modeling of the structuring element is based on the real data and implemented by parametric model approach. Several synthetic

• 전자공학회논문지 IE
• /
• 제46권4호
• /
• pp.57-65
• /
• 2009

본 논문에서는 대형 TFT-LCD 장치에서 소모전력 개선방안의 하나로 사용되는 Local Dimming 백라이트 제어방식을 휴대 단말기와 같은 소형 디스플레이에 적용하고자 이에 적합한 영상 컨트라스트 향상기법을 제안한다. 대형 디스플레이뿐만 아니라 소형 디스플레이 장치로 널리 각광받는 TFT-LCD의 광원인 백라이트 LED의 배치는 기존에 측면조사 구조로써 Global Dimming 방법이 사용되어졌으나 근래 진보된 기술을 통해 액정 후면에 배치, Local Dimming 방법이 제안되면서 입력영상을 블록단위로 분할 처리해야할 필요성이 대두되고 있다. 즉, 분할된 블록의 영상이 어두울 경우 해당 백라이트 LED의 공급전류를 감소시켜 명암대비를 극대화시킴과 동시에 소모전력량을 감소시키는 방법이다. 본 논문에서는 입력영상을 같은 크기의 블록단위로 분할한후 각 블록 내 화소정보를 분석하여 밝기성분(Y)에 대한 히스토그램과 평균, 표준편차 등의 정보를 획득하여 블록단위로 독립적인 컨트라스트 향상과정을 수행함으로써 영상정보를 통한 1차 컨트라스트 향상을 실현하고 동시에 블록단위 결과 정보를 Local Dimming을 조절하는 백라이트 제어부로 전달하여 광원제어를 통한 2차 컨트라스트 향상과 동시에 전류소모 감소 목표를 실현하고자 한다.