Research On Portable 3-D Scanning Technology Based On Time-Space Hybrid Structured-light Encoding

3-D model acquisition technology attracts much attention for its wide application in different circumstance. Portable 3-D scanning technology is a hot research topic in this field due to less restrictions and high speed.There are several ways to implement portable 3-D scanning such as ways based on structured-light,video or laser.However, none of these solutions is perfect for different reasons when being utilizedFirstly, there are technical and economic cost challenges.Space structured-light encoding method can be used to implment portable 3-D scanning but with low resolution.It is ralatively easy to acquire 3-D information by videos. But when scanning small objects,this method doesn't perform as well as scanning buildings. Laser 3-D scanner can provide high quality portable 3-D information acqusition with high speed. However,it's too expensitive for small research institute to buy a package of laser 3-D scanner and modeling software.Secondly, it's not easy to generate a complete 3-D model in single scanning process due to the reason of scanner's perspective range. By alignning models scanned form different perspective with softwares like RapidForm , the user is able to generate a complete model. It's time-consuming and inefficient.This dissertation attempted to develop a rapid structured -light 3-D acquisition system by introducing a time-space hybrid encoding method, which performed well portable 3-D scanning with high accuracy. The "rapid acquisition" means to increase the velocity of process which contains all procedures from acquiring images to displaying models.Four parts of contents were introduced in this dissertation.Firstly, In order to simplify human interaction and increase the efficiency of system calibration, an automatic calibration method for structured-light 3-D scanner was implemented to provide necessary parameters for triangulation when reconstructing models.Secondly, a time-space hybrid decoding method integrated high accuracy of time encoding method with high efficiency of space encoding method. A decoding algorithm with sub-pixel accuracy was developed and models were generated from different perspective by moving scanners.Thirdly, ICP registration algorithm was introduced to align these models from different views and to generate a complete 3-D model of moving objects.In the end, a prototype system was implemented. Hardware, software and main function were introduced in this dissertation.