• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.31 no.2
• /
• pp.159-164
• /
• 2013

It is well known that a single moving camera video is capable of extracting the 3-dimensional position of an object. With this in mind, current research performed image-based monitoring to establish a floating structure model using a camcorder system. Following this, the present study extracted frame images from digital camcorder video clips and matched the interest points to obtain relative 3D coordinates for both regular and irregular wave conditions. Then, the researchers evaluated the transformation accuracy of the modified SURF-based matching and image-based displacement estimation of the floating structure model in regular wave condition. For the regular wave condition, the wave generator's setting value was 3.0 sec and the cycle of the image-based displacement result was 2.993 sec. Taking into account mechanical error, these values can be considered as very similar. In terms of visual inspection, the researchers observed the shape of a regular wave in the 3-dimensional and 1-dimensional figures through the projection on X Y Z axis. In conclusion, it was possible to calculate the displacement of a floating structure module in near real-time using an average digital camcorder with 30fps video.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.12
• /
• pp.49-57
• /
• 2011

Extraction of regions of interest (ROIs) is the first and important step for the applications in computer vision and affects the rest of the application process. However, ROI extraction can be easily affected by the environment such as illumination, camera, etc. Many applications adopt problem-specific knowledge and/or post-processing to correct the error occurred in ROI extraction. In this paper, proposed is a robust framework that could overcome the environmental change and is independent from the rest of the process. The proposed framework uses a differential image and a color distribution to extract ROIs. The color distribution can be learned on-line, which make the framework to be robust to environmental change. Even more, the components of the framework are independent each other, which makes the framework flexible and extensible. The usefulness of the proposed framework is demonstrated with the application of hand region extraction in an image sequence.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.14 no.5
• /
• pp.73-79
• /
• 2014

In this paper, we proposed the mechanism which can measure traffic flow and vehicle speed on the highway as well as road by using the video and image processing to detect and track cars in a video sequence. The proposed mechanism uses the first few frames of the video stream to estimate the background image. The visual tracking system is a simple algorithm based on the sum of absolute frame difference. It subtracts the background from each video frame to produce foreground images. By thresholding and performing morphological closing on each foreground image, the proposed mechanism produces binary feature images, which are shown in the threshold window. By measuring the distance between the "first white line" mark and the "second white line"mark proceeding, it is possible to find the car's position. Average velocity is defined as the change in position of an object divided by the time over which the change takes place. The results of proposed mechanism agree well with the measured data, and view the results in real time.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.6
• /
• pp.28-34
• /
• 2020

This paper proposes an approach to the fusion of two heterogeneous sensors with two different fields-of-view (FOV): LIDAR and an RGB camera. Registration between data captured by LIDAR and an RGB camera provided the fusion results. Registration was completed once a depthmap corresponding to a 2-dimensional RGB image was generated. For this fusion, RPLIDAR-A3 (manufactured by Slamtec) and a general digital camera were used to acquire depth and image data, respectively. LIDAR sensor provided distance information between the sensor and objects in a scene nearby the sensor, and an RGB camera provided a 2-dimensional image with color information. Fusion of 2D image and depth information enabled us to achieve better performance with applications of object detection and tracking. For instance, automatic driver assistance systems, robotics or other systems that require visual information processing might find the work in this paper useful. Since the LIDAR only provides depth value, processing and generation of a depthmap that corresponds to an RGB image is recommended. To validate the proposed approach, experimental results are provided.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.1
• /
• pp.27-34
• /
• 2013

Background subtraction is one of the key techniques for automatic video content analysis, especially in the tasks of visual detection and tracking of moving object. In this paper, we present a new sample-based technique for background extraction that provides background image as well as background model. To handle both high-frequency and low-frequency events at the same time, multiple interval background models are adopted. The main innovation concerns the use of a confidence factor to select the best model from the multiple interval background models. To our knowledge, it is the first time that a confidence factor is used for merging several background models in the field of background extraction. Experimental results revealed that our approach based on multiple interval sampling works well in complicated situations containing various speed moving objects with environmental changes.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.5
• /
• pp.217-228
• /
• 2023

Video instance segmentation is an intelligent visual task with high complexity because it not only requires object instance segmentation for each image frame constituting a video, but also requires accurate tracking of instances throughout the frame sequence of the video. In special, human instance segmentation in drama videos has an unique characteristic that requires accurate tracking of several main characters interacting in various places and times. Also, it is also characterized by a kind of the class imbalance problem because there is a significant difference between the frequency of main characters and that of supporting or auxiliary characters in drama videos. In this paper, we introduce a new human instance datatset called MHIS, which is built upon drama videos, Miseang, and then propose a novel video data augmentation method, CDVA, in order to overcome the data imbalance problem between character classes. Different from the previous video data augmentation methods, the proposed CDVA generates more realistic augmented videos by deciding the optimal location within the background clip for a target human instance to be inserted with taking rich spatio-temporal context embedded in videos into account. Therefore, the proposed augmentation method, CDVA, can improve the performance of a deep neural network model for video instance segmentation. Conducting both quantitative and qualitative experiments using the MHIS dataset, we prove the usefulness and effectiveness of the proposed video data augmentation method.

• Korean Journal of Cognitive Science
• /
• v.25 no.3
• /
• pp.211-232
• /
• 2014

People search a certain object or a person so many time in a day. Besides the information about what the target is, perceptual information of the target can influence on the search process. In the current study, using an eye-tracker we aimed to examine whether the perceptual information of previewing target stimuli on the visual search process of the target and the task performance. Participants had to identify the previewing target stimulus presented in the middle of the screen, and then had to search the target among 8 items presented in a circle array, and had to decide whether the size of the target in the search display was same as that of the previewing stimulus. The experimental conditions were divided into 8 within-subject conditions by whether the search display was consisted of all the same size items or different size items (homogeneous search display vs. inhomogeneous search display), by the size of the preview target stimulus, and by the size of the target stimulus in the search display. Research hypothesis is that the size information of the previewing influence on the visual search process of the target and task performance when the items in the search display are in different sizes. In the results of behavioral data analysis, the reaction time showed the main effect of the search display, and the size of the target stimulus in the search display. and the interaction between the size consistency effect of target stimulus and the search display condition. In the results of analysis of eye-movement information, the Initial Saccade to Target Ratio measurement showed the interaction between the size consistency effect of target stimulus and the search display condition as the reaction time measurement did. That is, the size consistency effect of target stimulus only in the inhomogeneous search display condition indicated that participants searched the items in the same size as that of preview target stimulus. Post-hoc analyses revealed that the search and task performance in the inhomogeneous display condition were faster when the target size was consistent, but rather slower when the target size was inconsistent.

• Science of Emotion and Sensibility
• /
• v.18 no.4
• /
• pp.3-14
• /
• 2015

This study has extracted observation data by spatial factor for the analysis of subjects' selective attention with the objects of public space at the entrance of subway stations. The methods extracting observation data can be summarized as the following. First, the frequency analysis by lattice was prevalent for those methods, but there is a limitation to the analysis of the observation data. On the contrary, the method extracting observation data by factor applied in this study can make it clear if any sight is concentrated on any particular factors in a space. Second, the results from the extracted data corresponding to the observation area can be objectified while the method setting up the observation area by applying the radius of fovea. Third, time-sequential trace of observation results of relevant factors was possible through hourly analysis of spatial factors. The consideration of the results of "corresponding spatial scope" which is the object of this study will reveal that the more the observation time, the less the degree of attention it receives. Fourth, the frequency of observation superiority was applied for the analysis of the sections with selective attention by time scope; this revealed that men and women had intensive observation in time scope I (52.4 %) and in time scope IV (24.0 %), respectively.