블록 내 움직임 정보를 이용한 다단계 움직임 예측 알고리즘

곽내정,윤태승,한재혁,안재형 충북대학교 컴퓨터정보통신 연구소 2001 컴퓨터정보통신연구 Vol.9 No.2

본 논문에서는 불록 내의 움직임 정도에 따라 다단계의 볼록레벨을 설정하고, 그 블록레벨에 따라 각기 다른 다단계의 움직임 예측을 수행하는 다단계 움직임 예측 알고리즘을 제안한다. 제안방법은 3단계의 블륵 레벨을 설정하고. 각 불록 레벨에는 서로 다른 크기의 고정 블록을 설정한다 또한 각 블록 레벨에서 수행되는 움직임 예측에는 탐색 단계와 탐색 범위를 차등적으로 적용하였다. 이는 움직임이 없거나 적은 영역에서의 잉여 탐색 과정을 제거하여 계산적 효율성을 중대하고, 움직임 정도가 큰 영역에 대해서는 움직임 예측 과정을 심화시켜 예측의 정확성을 향상시킨다 실험 결과 프레임당 적은 수의 블록으로 고정된 크기의 볼록을 가진 전역 탐색 블록 정합 알고리즘보다 적은 예측 에러를 발생시켜 MSE를 20% 정도 감소시키며, 연산량을 95% 정도 절감하는 효과를 얻을 수 있었다. In this Paper. we Propose a multi-level block matching algorithm using motion information in blocks. In the Proposed algorithm, the block-level of a block is decided by the motion degree in the block before motion searching Procedure, and then adequate motion searching performs according to the block-level. The number of block-level is 3 step and different fixed-block sixte is assigned to each block-level. Also. in motion estimation of each block-level is differential applied search step and search area. Which imProves computational efficiency by eliminating an unnecessary searching Process in no motion or low motion regions. and brings more accurate estimation results by deepening motion searching process in high motien legions. Simulation results shew that the proposed algorithm brings the lower estimation error about 20% MSE(mean squared error) reduction with the fewer blocks per frame and the lower computational loading about 98% operational amount reduction than full search block matching algorithm with constant block size.

이진 디더링 이미지에 적용하는 디지털 워터마킹의 성능개선

김현주,곽내정,안재형 충북대학교 컴퓨터정보통신 연구소 2001 컴퓨터정보통신연구 Vol.9 No.1

본 논문에서는 이진 프린팅 영상에 워터마크를 삽입하는 방법을 제안한다. 제안방법은 영상을 디더링 한 후 16×16 크기의 블록으로 나누어 워터마크 값(0 or 1)에 따라 다르게 정의된 기준 블록 패턴과 비교함으로 워터마크를 삽입한다. 워터마크 검출은 워터마크가 삽입된 영상의 카운팅 배열을 생성하고 생성된 카운팅 배열로 워터마크 배열을 만들어 디더링 영상에 삽입된 워터마크 배열과 비교함으로 된다. 성능평가는 워터마크 삽입영상을 프린트 한 후 스캔하여 기존의 방법과 워터마크 검출 성능을 비교하였으며 제안 방법은 기존 방법보다 3~5% 정도 워터마크 검출성능이 개선되어졌다. In this paper we propose a novel method which watermarks are embedded in printed images by exploiting the printing process. First, we convert a original image to a dithered binary image and a dithered binary image divided into the blocks of 16×16 pixel. And then watermark is embedded in a dithered binary image by comparing that blocks with reference block patterns which are determined by watermark values(0 or 1). We extract watermark by comparing the watermark which is reconstructed from the watermark embedded image with the original watermark which is embedded in a binary image. To examine we compare performance of the proposed algorithm with that of the conventional watermark embedding algorithm for printed images by detecting watermark for scan images. Simulation result shows that the proposed algorithm works better than the conventional algorithm.

움직임 예측을 위한 비트패턴을 이용한 고속 워핑 예측

강봉구,윤태승,곽내정,안재형 충북대학교 컴퓨터정보통신 연구소 2002 컴퓨터정보통신연구 Vol.10 No.2

본 논문에서는 비트패턴을 이용한 고속 워핑 예측을 제안한다. 워핑 예측은 이웃한 노드들의 움직임 벡터에서 움직임 정보를 가져오는 공간적인 의존성 때문에 최적의 움직임을 찾기 위해서는 반복적인 탐색을 요구한다. 반복적인 탐색에 의해 발생하는 연산량의 증가는 워핑 예측의 사용을 저해하는 큰 이유 중의 하나이다. 본 논문에서 제안한 방법은 비트패턴을 이용해서 움직임 여부를 판단한다. 움직임이 없는 영역에서의 의미 없는 움직임 예측을 제거하고, 움직임이 있는 영역에서만 워핑 예측을 수행함으로써 빠른 움직임 예측이 가능하게 된다. 제안된 방법으로 실험한 결과 기존의 워핑 예측을 수행했을 때보다 성능은 근접하면서도 연산량이 75% 이상 감소하였다. In this paper, we propose a fast warping prediction using bit-pattern. Because of the spatial dependency between motion vectors of neighboring node points carrying motion information, the optimization of motion search requires an iterative process. The computational load stemming from the iterative search is one of the major obstacles for practical usage of warping prediction. The motion estimation in the proposed algorithm measures whether the motion content of the area is or not, using bit-pattern. Warping prediction using the motion content of the area make the procedure of motion estimation efficient by eliminating an unnecessary searching process. Experimental results show that the proposed algorithm can reduce about 75% of iterative search while maintaining performance as close as the conventional warping prediction.

The Notification System of Dangerous Situation using the Change of Facial Expression

Nae-Joung Kwak,Teuk Seob Song 한국정보통신학회 2015 2016 INTERNATIONAL CONFERENCE Vol.7 No.1

In this paper, we propose an algorithm to determine a risk situation with the change of facial expressions, and to service information that notifies the danger situation. The proposed method defines fear and surprise among the various human emotions as human feeling in dangerous situations, and recognizes two facial expressions. The proposed method firstly converts the input image into a gray image, extracts the face area of the object of interest, and extracts the eye region and lip region from the detected face region. After then it apply Uniform LBP for each region to detect facial expression. Facial expression is firstly detected the expression of fear and surprise in the mouth area, and finally recognize the facial expressions of fear and surprise by the characteristics of the eye area. If the expressions of fear and surprise are continuous, the proposed algorithm defines that situation as dangerous situation and sends status information to smartphones. The proposed method is tested in a number of images of MUCT database used for facial expression recognition and the input video image from a web cam. Facial expression recognition rate by the proposed method is 91%. It was a good result in detecting dangerous situations and alerting danger information.

Human Action Recognition Using Accumulated Moving Information

Nae-Joung Kwak,Teuk-Seob Song 보안공학연구지원센터 2015 International Journal of Multimedia and Ubiquitous Vol.10 No.10

This paper proposes a human action recognition algorithm which can be efficiently applied to a real-time intelligent surveillance system. This method models the background, obtains the difference image between input image and the modeled background image, extracts the silhouette of human object from input image, and recognizes human action by using coordinates of object, directions of that and accumulated moving regions of that. The human actions recognized in this study amount to a total of 8 type of actions, which include walking, raising an arm (left, right), raising a leg (left, right), sitting and crouching. The proposed method has been experimented for 8 different movements using 4 people using video input of a webcam and it has shown good results in terms of recognizing human action.

Quality Improvement of Interpolated Image Using Weight-Granting Method Based Median Values Of Local area

Nae Joung Kwak 한국콘텐츠학회 2008 ICCC International Digital Design Invitation Exhib Vol.2008 No.12

Human Action Recognition for Risk Notification Service

Nae-Joung Kwak,Teuk-Seob Song 한국정보통신학회 2014 2016 INTERNATIONAL CONFERENCE Vol.6 No.1

This paper proposes a human action recognition algorithm which can be efficiently applied to a real-time intelligent surveillance system. This method models the background image or video stream, obtains the difference image between input image and the modeled background image, extracts the silhouette of human object from input the image, and recognizes human actions. In order to recognize human actions, the proposed method uses direction vector of movement and link codes of movement dependent histograms (LCMDH). Human actions are classified into walking, sitting, standing up, and unusual actions (falling down and fainting. According to the test results the proposed algorithm efficiently classified human actions, detected unusual actions, and provided an alarm service.

Automatic Detection Algorithm Of Joint Point Of Human Body For Avatar Animation

Nae-Joung Kwak,Teuk-Seob Song 한국콘텐츠학회 2010 한국콘텐츠학회 종합학술대회 논문집 Vol.2010 No.12

Human Action Recognition Algorithm: Risk Notification Service on the Android Environments

Nae-Joung Kwak,Teuk-Seob Song 보안공학연구지원센터 2014 International Journal of Smart Home Vol.8 No.5

Joint Tracking and Transmission System for Simulating Motion of the Human Body on Android Smart Phone

Nae Joung Kwak,Teuk-Seob Song 보안공학연구지원센터(IJCA) 2011 International Journal of Control and Automation Vol.4 No.4