Using two color checkerboards for image based 3D stair reconstruction

• Main Authors: Chung-Chien Lan, 藍中謙
• Other Authors: Tzung-Han Lin
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/91180108883393746094

碩士 === 國立臺灣科技大學 === 色彩與照明科技研究所 === 104 === Population aging and the building’s age becoming older are the important issues for people’s well-being. So the home safety problem will become a key link. To purchase a new house in Metropolitan now is an almost impossible job. And remodeling an old house is also a difficult mission. So the compact Aids (Assistance for individuals with disabilities) is more than best choice to setup in old house. The most famous product is stairlift chair. However, the lift chair is a highly customized product. In the first step, it has to build the stair including 3D digital reconstruction in client’s house. Second step is using the 3D model to plan and design the route for the stairlift chair. Nowadays, most suppliers still use the traditional method to measure the real size of the stairs and to build 3D models by 3D CAD software. Thus, it wastes time and human resource. Accordingly, finding an effective and easy practical solution is critical to improve the process. First of all, we perform a 3D reconstruction solution in dealing with design of the route for stairlifts. We use the vision-based 3D reconstruction due to its efficient productivity. And we use the two color chessboard for determining the intrinsic and extrinsic parameters of the camera, which means the calibration is carried out in each time. After getting the key parameters of the camera, we can use the coordinates from extrinsic parameters and the homography matrix, to transform the corners of chessboards into the Euclidean space. After obtaining all corner features from all images, we can alignment the points. Finally, we will have 3D stair reconstruction model. The two-color-chess checkerboard system is built based on both addictive and subtractive color methods. Because the camera can record the data with three color channels, says red, green and blue. So, we find a color channel which is able to readily identify the cyan and magenta chess checkerboards. Transforming the image into other color space is used to find a perfect color channel and the to get a clear separated picture. Finally, by following the result of the separating color test, we will find a new evaluation process to be a standard for verifying available checkerboards. Thus, it can assist us to evaluate the samples that can be used in the reconstruction system or not.