• International Journal of Contents
• /
• 제8권1호
• /
• pp.16-22
• /
• 2012

In this paper we propose a method for verifying video in a video telephony system implemented in DM6446 DaVinci Processor. Each frame is categorized either error free frame or error frame depending on the predefined criteria. Human face is chosen as a basic means for authenticating the video frame. Skin color based algorithm is implemented for detecting the face in the video frame. The video frame is classified as error free frame if there is single face object with clear view of facial features (eyes, nose, mouth etc.) and the background of the image frame is not different then the predefined background, otherwise it will be classified as error frame. We also implemented the image histogram based NCC (Normalized Cross Correlation) comparison for video verification to speed up the system. The experimental result shows that the system is able to classify frames with 90.83% of accuracy.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2004년도 ICCAS
• /
• pp.1235-1238
• /
• 2004

Thresholding has been used to reduce the number of gray values in images. Typically, a single threshold value has been used, resulting in two gray level images. Image reduction of one single threshold value, however, may lose too much of the high-frequency edge information. Thus, dynamic thresholding that uses a different threshold for each pixel is preferred instead of using a single threshold value. Dynamic thresholding can preserve high frequency details as well as reduce the size of images. Since it takes long time to perform existing software dynamic thresholding in an embedded system, this paper proposes and implements a circuit by using a FPGA in order to perform a real-time dynamic thresholding,. The proposed circuit consists of two counters, and threshold look-up table, and control unit. The values of two counters determine each pixel position, the threshold look-up table converts each pixel value into other value, and the control unit generates necessary control signals. On arriving from a camera to the proposed circuit, each pixel is compared with its threshold value and is converted into other gray value. An image processing system by using the proposed circuit will be implemented and some experiments will be performed.

• 한국측량학회지
• /
• 제35권5호
• /
• pp.339-348
• /
• 2017

Due to recent improvements in computer processing speed and image processing technology, researches are being actively carried out to combine information from a camera with existing GNSS (Global Navigation Satellite System) and dead reckoning. In this study, the mathematical model based on SPR (Single Photo Resection) is derived for image-based assistant algorithm for vehicle positioning. Simulation test is performed to analyze factors affecting SPR. In addition, GNSS/on-board vehicle sensor/image based positioning algorithm is developed by combining image-based positioning algorithm with existing positioning algorithm. The performance of the integrated algorithm is evaluated by the actual driving test and landmark's position data, which is required to perform SPR, based on simulation. The precision of the horizontal position error is 1.79m in the case of the existing positioning algorithm, and that of the integrated positioning algorithm is 0.12m at the points where SPR is performed. In future research, it is necessary to develop an optimized algorithm based on the actual landmark's position data.

• Journal of Multimedia Information System
• /
• 제8권4호
• /
• pp.203-210
• /
• 2021

Deep leaning convolutional neural networks (CNN) have successfully been applied to image super-resolution (SR). Despite their great performances, SR techniques tend to focus on a certain upscale factor when training a particular model. Algorithms for single model multi-scale networks can easily be constructed if images are upscaled prior to input, but sub-pixel convolution upsampling works differently for each scale factor. Recent SR methods employ multi-scale and multi-path learning as a solution. However, this causes unshared parameters and unbalanced parameter distribution across various scale factors. We present a multi-scale single-path upsample module as a solution by exploiting the advantages of sub-pixel convolution and interpolation algorithms. The proposed model employs sub-pixel convolution for the highest scale factor among the learning upscale factors, and then utilize 1-dimension interpolation, compressing the learned features on the channel axis to match the desired output image size. Experiments are performed for the single-path upsample module, and compared to the multi-path upsample module. Based on the experimental results, the proposed algorithm reduces the upsample module's parameters by 24% and presents slightly to better performance compared to the previous algorithm.

• 한국가금학회지
• /
• 제22권4호
• /
• pp.233-238
• /
• 1995

A computer vision system was built to generate images of a single, stationary egg. This system includes a CGD camera, a frame grabber, and incandescent back lighting system. Image processing algorithms were developed to inspect egg shell and to sort eggs. Those values of both gray level and area of dark spots in the egg image were used as criteria to detect holes in egg and those values of both area and roundness of dark spots in the egg image were used to detect cracks in egg. For a sample of 300 eggs, this system was able to correctly analyze an egg for the presence of a defect 97.5% of the time. The weights of eggs were found to be linear to both the projected area and the perimeter of eggs viewed from above. Those two values were used as criteria to sort eggs. The coefficients of determination(r$^2$) for the regression equations between weights and those two values were 0.967 and 0.972 in the two sets of experiment. Accuracies in grading were found to be 95.6% and 96.7% as compared with results from sizing by electronic weight scale.

• 한국정보통신학회논문지
• /
• 제12권8호
• /
• pp.1481-1487
• /
• 2008

본 논문에서는3D환경에서의 사용을 위한 새로운 형태의 인터렉티브 3D 플로팅 영상 포인팅 장치를 제안한다. 이 장치는 플로팅 렌즈 배열을 사용한3D플로팅 영상 재생 시스템과 실시간 손가락 검출 기반의 버턴형 인터페이스로 구성된다. 제안하는 장치에서는 사용자가 손가락을 이용하여 3D 공간에 플로팅 된 영상 배열에서 원하는 영역을 지시하고, 2대의 카메라를 이용하여 사용자의 손가락 인식하여 원하는 이벤트를 발생하는 인터렉션 기능을 효과적으로 수행한다. 제안하는 시스템의 유용함을 보이기 위해서, 실험적인 장치 구현을 수행하고 기초적인 실험 결과를 보고한다.

• Journal of Information Processing Systems
• /
• 제18권1호
• /
• pp.59-74
• /
• 2022

During the last decade, the security of digital images has received considerable attention in various multimedia transmission schemes. However, many current cryptosystems tend to adopt a single-layer permutation or diffusion algorithm, resulting in inadequate security. A hierarchical bilateral diffusion architecture for color image encryption is proposed in response to this issue, based on a hyperchaotic system and DNA sequence operation. Primarily, two hyperchaotic systems are adopted and combined with cipher matrixes generation algorithm to overcome exhaustive attacks. Further, the proposed architecture involves designing pixelpermutation, pixel-diffusion, and DNA (deoxyribonucleic acid) based block-diffusion algorithm, considering system security and transmission efficiency. The pixel-permutation aims to reduce the correlation of adjacent pixels and provide excellent initial conditions for subsequent diffusion procedures, while the diffusion architecture confuses the image matrix in a bilateral direction with ultra-low power consumption. The proposed system achieves preferable number of pixel change rate (NPCR) and unified average changing intensity (UACI) of 99.61% and 33.46%, and a lower encryption time of 3.30 seconds, which performs better than some current image encryption algorithms. The simulated results and security analysis demonstrate that the proposed mechanism can resist various potential attacks with comparatively low computational time consumption.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2007년도 Proceedings of ISRS 2007
• /
• pp.625-628
• /
• 2007

We present an efficient and robust technique to estimate the velocity of moving targets from a single SAR image. In SAR images, azimuth image shift is a well known phenomenon, which is observed in moving targets having slant-range velocity. Most methods estimated the velocity of moving targets from the distance difference between the road and moving targets or between ship and the ship wake. However, the methods could not be always applied to moving targets because it is difficult to find the road and the ship wake. We adopted a method estimating the velocity of moving targets from azimuth differentials of range-compressed image. This method is based on an assumption that Doppler center frequency shift of moving target causes a phase difference in azimuth differential values. The phase difference is linearly distorted by Doppler rate due to the geometry of SAR image. The linear distortion is eliminated from phase removal procedure, and the constant phase difference is estimated. Finally, range velocity estimates for moving targets are retrieved. This technique is tested using an ENVISAT ASAR image in which several unknown ships are presented. The theoretical accuracy of this technique is discussed by SAR simulation. The advantages and disadvantages of this method over the conventional method are also discussed.

• 한국산학기술학회논문지
• /
• 제6권4호
• /
• pp.349-354
• /
• 2005

컴퓨터 비전에 관한 연구는 인간의 시각 능력을 대신할 시스템을 구축하고자 하는 것으로 특히 이동 객체추적 시스템은 최근 중요한 연구 분야로 대두되고 있다. 본 연구에서는 움직임 추정에 기반 한 단일 이동객체를 추적할 수 있는 시스템을 제안하였다. 추적 시스템은 차영상 기법을 이용하여 객체의 움직임을 추정하곤 카메라의 Pan/Tilt를 제어함으로서 이동 객체를 추적할 수 있도록 하였다. 구현된 시스템은 영상획득 및 전처리 단계, 움직임 추정 단계, 객체 추적 단계로 구성하였다. 실시간 추적 실험을 실시한 결과 이동 객체의 움직임을 추정할 수 있었고, 추적 과정에서도 객체를 일어버리지 앉고 정상적으로 추적할 수 있었다.

• 방송공학회논문지
• /
• 제23권3호
• /
• pp.369-382
• /
• 2018

본 논문은 딥 컨볼루션 뉴럴 네트워크(CNN)를 이용하여 행 별로 서로 다른 노출로 촬영된 단일 영상을 HDR 영상으로 변환하는 기법을 제안한다. 제안하는 알고리즘은 먼저 입력 영상에서 저조도 또는 포화로 인해 발생하는 정보 손실 영역을 CNN을 이용하여 복원하여 휘도맵을 생성한다. 또한, CNN 학습 과정에서 인간의 시각 인지 특성을 고려할 수 있는 손실 함수를 제안한다. 마지막으로 복원된 휘도맵에 디모자이킹 필터를 적용하여 최종 HDR 영상을 획득한다. 컴퓨터 모의실험을 통해 제안하는 알고리즘이 기존의 기법에 비해서 높은 품질의 HDR 영상을 취득하는 것을 확인한다.