• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2014.06a
• /
• pp.28-31
• /
• 2014

본 논문에서는 블랙박스 혹은 운전석에 장착된 카메라로부터 얻어진 차량 영상에 대한 영역별 수직 히스토그램 매칭 및 선형 회귀분석 모델(linear regression model)을 활용한 강건한 차량 운행 동영상의 안정화(video stabilization) 기법을 제안한다. 동영상 안정화 기법은 영상의 흔들림 보정뿐 아니라 동영상 내 강건한 특징점 추적 및 매칭을 위한 이전의 전처리 과정으로 적용된다. 일반적으로 촬영 과정에서 많은 떨림이 포함될 수 있는 야외 CCTV 영상이나 손으로 들고(hand-held) 촬영된 동영상에 대한 흔들림 보정 등에 적용되고 있으나 영상 내 특징점이 지속적으로 변하고 영상의 변화 정도가 매우 심한 차량 운행 동영상에서는 적용된 사례가 드물다. 본 연구에서는 일반적인 비디오 안정화 기술이 적용되기 어려운 차량 운행 동영상에 대하여 수직 투영 히스토그램 매칭 및 선형 회귀분석 모델 기반의 안정화 기법을 제안한다. 제안된 기법은 입력영상에 대한 영역별 수직 투영 히스토그램 매칭을 수행하고 선형 회귀모델을 통해 영상에 나타나는 수직 및 회전이동 변환을 선형 근사하여 시간 영역 상의 입력 영상에 대한 안정화를 달성한다. 제안 방법의 검증을 위해 블랙박스로 촬영된 실제 동영상에 동영상 안정화 기술을 적용하였으며, 운행 중 불규칙한 노면으로 인한 영상의 흔들림이 효과적으로 제거되는 것을 확인할 수 있었다.

• Journal of Intelligence and Information Systems
• /
• v.29 no.2
• /
• pp.115-127
• /
• 2023

Video stabilization is one of the camera technologies that the importance is gradually increasing as the personal media market has recently become huge. For deep learning-based video stabilization, existing methods collect pairs of video datas before and after stabilization, but it takes a lot of time and effort to create synchronized datas. Recently, to solve this problem, unsupervised learning method using only unstable video data has been proposed. In this paper, we propose a network structure that learns the stabilized trajectory only with the unstable video image without the pair of unstable and stable video pair using the Convolutional Auto Encoder structure, one of the unsupervised learning methods. Optical flow data is used as network input and output, and optical flow data was mapped into grid units to simplify the network and minimize noise. In addition, to generate a stabilized trajectory with an unsupervised learning method, we define the loss function that smoothing the input optical flow data. And through comparison of the results, we confirmed that the network is learned as intended by the loss function.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.37 no.4
• /
• pp.399-405
• /
• 2009

The focal plane image stabilization for a satellite camera is one of the an effective method which can increase the satellite camera's image quality by removing the motion disturbance of a focal plane. The objectives of this article are to introduce the concept of the focal plane image stabilization and determine the best driving conditions of the actuator for the response and thrust. Under various driving condition the experiments have been performed to investigate the response and thrust characteristics of the piezoelectric driven stick-slip actuator of the focal plane image stabilizing device. From experiments, the best driving frequency and duty ratio for the magnesium slider are 70 kHz and 27%, respectively.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.39 no.2
• /
• pp.227-233
• /
• 2015

In this study, an indoor test environment was developed for studying the vision stabilizer of an unmanned vehicle, using a virtual environment and a 6-axis motion simulator. The real driving environment was replaced by a virtual environment based on the Aberdeen Proving Ground bump test course for military tank testing. The vehicle motion was reproduced by a 6-axis motion simulator. Virtual reality driving courses were displayed in front of the vision stabilizer, which was located on the top of the motion simulator. The performance of the stabilizer was investigated by checking the image of the camera, and the pitch and roll angles of the stabilizer captured by the IMU sensor of the camera.

• Journal of Broadcast Engineering
• /
• v.19 no.6
• /
• pp.798-809
• /
• 2014

Video stabilization is performed to remove unexpected shaky and irregular motion from a video. It is often used as preprocessing for robust feature tracking and matching in video. Typical video stabilization algorithms are developed to compensate motion from surveillance video or outdoor recordings that are captured by a hand-help camera. However, since the vehicle video contains rapid change of motion and local features, typical video stabilization algorithms are hard to be applied as it is. In this paper, we propose a novel approach to compensate shaky and irregular motion in vehicle video using linear regression model and vertical projection histogram matching. Towards this goal, we perform vertical projection histogram matching at each sub region of an input frame, and then we generate linear regression model to extract vertical translation and rotation parameters with estimated regional vertical movement vector. Multiple binarization with sub-region analysis for generating the linear regression model is effective to typical recording environments where occur rapid change of motion and local features. We demonstrated the effectiveness of our approach on blackbox videos and showed that employing the linear regression model achieved robust estimation of motion parameters and generated stabilized video in full automatic manner.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.15 no.4
• /
• pp.957-964
• /
• 2011

The image stabilization is the procedure of stabilizing the blurred image with image processing method. Due to easy detection of global motion, PA(Projection algorithm) based on digital image stabilization has been studied by many researchers. PA has the advantage of easy implementation and low complexity, but in the case of serious rotational motion the accuracy of the algorithm will be cut down because of its fixed exploring range, and, on the other hand, if extending the exploring range, the block for detecting motion will become small, then we cannot detect correct global motion. In this paper, to overcome the drawback of conventional PA, an Iterative Projection Algorithm (IPA) is proposed, which improved the correctness of global motion by detecting global motion with detecting block which is appropriate to different extent of motion. With IPA, in the case of processing 1000 continual frames shot in automobile, compared with conventional algorithm and other detecting range, the results of PSNR is improved 6.8% at least, and 28.9% at the most.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.19 no.1
• /
• pp.145-152
• /
• 2019

In this paper, we proposed a shaking image stabilization algorithm using deep learning. The proposed algorithm utilizes deep learning, unlike some 2D, 2.5D and 3D based stabilization techniques. The proposed algorithm is an algorithm that extracts and compares features of shaky images through CNN network structure and LSTM network structure, and transforms images in reverse order of movement size and direction of feature points through the difference of feature point between previous frame and current frame. The algorithm for stabilizing the shake is implemented by using CNN network and LSTM structure using Tensorflow for feature extraction and comparison of each frame. Image stabilization is implemented by using OpenCV open source. Experimental results show that the proposed algorithm can be used to stabilize the camera shake stability in the up, down, left, and right shaking images.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.47 no.6
• /
• pp.10-18
• /
• 2010

This paper proposes an image stabilization algorithm for close watching UAV(Unmanned Aerial Vehicle) using motion separation and stabilization mode. The motion of UAV is composed of its actual navigating motion and unwanted vibrating motion so that image sequences obtained from UAV are shaken randomly. In order to stabilize these images we separate the vibrating motion component from UAV motion and remove the effect caused by it from image sequences. In the proposed algorithm the motion and global intensity change of two consecutive images are modeled with 6 motion parameters and 2 intensity change parameters respectively. These modeled parameters are estimated by non-linear least square method based on Gauss-Newton algorithm. The vibrating motion component is separated from the estimated motion using IIR filtering and the geometric deformation caused by it is removed from image sequences. In order to apply the proposed method to real aerial image sequences with many abrupt changes of camera view, we proposed a stabilizing method using two different modes named as stabilizing and non-stabilizing mode. Experimental results show that the accuracy of motion estimation is 99% and the efficiency of removing the vibrating motion component is 90%. We apply the proposed method to real aerial image sequences and verified its stabilizing performance.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.19 no.8
• /
• pp.1805-1810
• /
• 2015

Increasing of depth data accessibility, depth data is used in many researches. Motion recognition of computer vision also widely use depth image. More accuracy motion recognition system needs more stable depth data. But depth sensor has a noise. This noise affect accuracy of the motion recognition system, we should noise suppression. In this paper, we propose using spatial domain and temporal domain stabilization for depth image and makes it hardware IP. We adapted our hardware to floor removing algorithm and verification its effect. we did realtime verification using FPGA and APU. Designed hardware has maximum frequency 202.184MHz.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2017.11a
• /
• pp.204-206
• /
• 2017

최근에 개인용 카메라를 통해 개인의 추억을 파노라마 영상으로 기록하는 것에 관심이 급증하고 있다. 파노라마 영상에 관심이 급증함에 따라 파노라마 영상을 제작하는 방법에 대해 여러 분야에서 연구가 많이 진행되고 있다. 일반적으로 개인용 카메라를 손으로 잡고 촬영하는 경우가 대부분이다. 손으로 잡고 촬영한 영상은 손 떨림에 의해 흔들린 영상이 된다. 이는 파노라마 영상을 만들 때 어려운 요소를 야기한다. 그러므로 흔들린 영상을 정합하고 안정화하는 연구는 매우 중요하다. 따라서 본 연구의 목적은 최근에 연구된 비디오 정합(Video Stitching)과 비디오 안정화(Video Stabilization)의 정확도 및 경향을 파악을 통해 빛의 변화가 빈번하고 움직임이 많은 콘서트 영상 정합에 이용될 아이디어 추출에 있다.