Measuring and Visualizing Geometrical Differences Using a Consumer Grade Range Camera
Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Measuring geometrically complex parts is currently very labor and time intensive in most cases because it involves a significant amount of manual steps to perform a correct measurement. This project was done to find a more productive method of measuring these parts in a production environment. The method should be more productive in terms of speed and ease of use, by automation large parts of the process. The goal is to develop a solution that can measure and check parts automatically using a range camera. A range camera is selected because the data acquisition speed is significantly higher than that of a coordinate measuring machine (CMM). The only manual step in this process is the actual scanning, the remaining model making and comparison steps are fully automatic. A two-step method is presented that shows that it is possible to measure parts in a more productive manner, than conventional techniques used to measure object with a complex geometry. The developed method takes the depth images from the range camera and turns them into a point cloud model of the object that needs to be measured. The second step of the solution then automatically compares the point cloud against a CAD-model of the object. This second step is the key to improve measurement productivity. The end result of this is a proof of concept program that incorporates the two step method, that takes in range camera images and as output it gives the scanned object measured against a reference model. The labor time is reduced to the scanning time which is less than a minute. The total processing time to make the comparison takes roughly fifteen minutes depending on the amount of scanning data. The processing can be done in parallel to other production steps. Because a consumer grade range camera is used, only centimeter accuracy is reached. Using more advanced range cameras or similar techniques will lead to more accurate results as noise in the image, is the limiting factor for the algorithm developed here.
Place, publisher, year, edition, pages
2015. , 46 p.
, Degree Project in Production Engineering Management, Second Level, 636
Engineering and Technology
IdentifiersURN: urn:nbn:se:kth:diva-167357OAI: oai:DiVA.org:kth-167357DiVA: diva2:813236