Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/91304
標題: The Alignment of Range sensor and Rotating table
深度感測器及轉盤之校準
作者: 廖晉逸
Jin Yi Liao
關鍵字: Range sensor
Kinect RANSAC
least square method
shaft locating
Range sensor
Kinect
隨機抽樣一致演算法
最小平方法
轉軸定位
引用: 1.Philip J. Noonan, William A. Hallettt, 'The Design and Initial alibration of an Optical Tracking System Using the Microsoft Kinect' 2.Victor Sanchez, Avideh Zakhor 'PLANAR 3D MODELING OF BUILDING INTERIORS FROM POINT CLOUD DATA' 3.Tanwi Mallick, Partha Pratim Das, and Arun Kumar Majumdar 'Characterizations of Noise in Kinect Depth Images' 4.Arshad Jamal, Praveen Mishra, Subrata Rakshit 'Real-time Ground Plane Segmentation and Obstacle Detection for Mobile Robot Navigation' 5.Matthias Hernandez, Jongmoo Choi and G'erard Medioni 'LASER SCAN QUALITY 3-D FACE MODELING USING A LOW-COST DEPTH CAMERA' 6.Li Shibo, Zhuo Qing 'A New Approach to Calibrate Range Image and Color Image From Kinect' 7.Masahiro Tanaka 'Robust parameter estimation of road condition by Kinect sensor' 8.Stefan Oßwald, Jens-Steffen Gutmann, Armin Hornung, Maren Bennewitz 'From 3D Point Clouds to Climbing Stairs:A Comparison of Plane Segmentation Approaches for Humanoids'
摘要: In measuring the shape of object, if we want faster but relative inaccurate answer , we will choose contactless way such as infrared camera to get the images in different angle and register it in one. However, the current registration method needs good initial values to shorten the time for the iterative. So, in this paper we present a new method to locate the turntable's spindle orientation and position, in order to calibrate or registration. To achieve our goal, we design a hexagonal column for detection. By repeatedly measure the hexagonal column, we can find and filter out the edge points, and use the RANSAC and least square method to locate the precise position of the hexagonal column, moreover, use our designed algorithm to locate the turntable's spindle orientation and position. The result proves our calculated turntable's spindle orientation and position are quite precision. After all the measurement, we will discuss the issue which might affect the accuracy in two parts. First we discuss about the factor that might affect the accuracy of the hexagonal column, such as the distance between Kinect and the hexagonal column, surface texture, RANSAC algorithm's counting times. The second part is about relationship between the hexagonal column and the turntable, such as the distance between the hexagonal column and the turntable's middle axis, the inclination, and the number of the images uses in the algorithm.
於量測物體外形時,若希望採取快速但相對來說較不精準的方式,則我們會選取以非接觸式,即以紅外線攝影機取得物體之外型輪廓並以配準方式組合各方向所拍得之影像,但現行的多種配準法皆需要良好的初始值以縮短其迭代所需時間,因此本文提出了一方法,用以快速於空間中定位出轉盤之方向及位置,用於後續之校正、配準等計算 為了達到我們的目的,我們設計了一個六角柱標準件用於偵測,藉由多次量測此六角柱,找出空間中之邊緣點,進一步篩選後,使用隨機抽樣一致演算法及最小平方法找出準確之六角柱位置,並以我們所提出之演算法進一步求出六角柱放置之平面及所旋轉之方向並比較之,結果證實我們所得出之轉盤平面及旋轉方向皆相當正確 於量測完成後,我們也會針對各種於演算法中可能會影響精準度之要素分兩部分討論,頭一部分討論關於可能影響六角柱之精準度之要素,如Kinect和標準件之間之距離、表面材質,隨機抽樣一致演算法之演算次數等等,第二部分則討論關於標準件和轉盤之間的關系會造成之差異,如標準件之位置偏離主軸距離,傾斜與否,所使用之標準件影像張數差異等等。
URI: http://hdl.handle.net/11455/91304
文章公開時間: 2015-08-28
Appears in Collections:機械工程學系所

文件中的檔案:

取得全文請前往華藝線上圖書館



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.