Paper Title
Image Analysis of Modular Units using Computer Vision

Abstract
Recently the Korea ministry of education has been introducing ways to expand public kindergartens using modular construction method as a solution for private kindergarten closure. Since a modular architecture has many advantages, which are minimizing on-site assembly through factory production and impact of climate, mass-produced, easy to assemble, shortening the construction period and moving or re-using buildings. However, modular construction is mainly a prefabricated building using wood or steel frame, causing vibration problems due to manufacturing and construction errors. There is nothing more important to reduce the error rate at manufacturing because the modular building has high production rate in manufacture state. Modular architecture is defined in various fields such as planning, de-sign, module production, field construction, operation and maintenance. However, in this research, it is the main purpose to improve precise workability by minimizing an error rate due to displacement of members that may occur during the manufacturing, moving and constructing. In recent years, researchers from overseas researchers have been interested in three areas: construction management and lean construction projects conducted by Louisiana State University researchers in the United States, and construction management and lean construction in Alberta, Canada. Building Information Modeling(BIM) and production management from the university's research team and the High-rise modular building construction from the SCI research team in the UK. In the UK SCI's paper on high-rise modular architecture, (Wolverhampton Victoria Hall), UK, the contents of the high-rise residential modular construction method, which is a major concern of domestic modular construction companies, are covered. It is expected that research on modular high-rise technology will continue. The existing study (middle-floor modular research group) includes joint method development considering construction and manufacturing error and development of displacement control technology during construction. However, this study aims at developing image sensing element technology through image analysis, and it can be applied to both image sensing and image sensing. There is a difference in the development of precision improvement technology utilized. In this study, we will reconstruct the 2D image into 3D through camera calibration and calculate the error angle generated at the construction stage of the modular building using the image processing technique. 3D reconstruction was performed using Matlab Computer Vision Toolbox. The reference image was taken and experimented with 8 cases with angles of 0.5〬, 1〬, 1.5〬, 2〬, 3〬, 5〬, 7〬, 10〬. And images for calibration 4, 8, 16, 64, 128 images and images in a concentrated and distributed type. Then an analysis of each case was performed. Prior to this experiment, the reliability of the calibration was verified by comparing the images taken with the same frame, and the results showed the highest accuracy at the center distance of the distribution of calibration images. Except for 5〬, the overall error rate less than 3% showed similar estimation results. Keywords - Computer Vision, Camera Calibration, 3D, Modular, Image Processing