会议专题

Complex Object 3D Measurement Based on Neural Network and Digital Fringe Projection

Purpose The purpose of this paper is to report on a study of the accurate phase-height mapping algorithm in the structured light system based on digital fringe projection.Design/methodology/approach Neural network training is a well known method to approximate a nonlinear system without an explicit physical model. In this work, a novel accurate phase-height mapping algorithm based on neural network is proposed. A group phase shift fringe patterns were projected onto a calibration board with 99 circle calibration marks from different heights and captured by the CCD camera.Then high accurate training sample data sets can be extracted from the images. By training the network, the relationship between the images coordinates and the 3D coordinates of the object can be obtained.Findings In support of this work, 3D accuracy measurements of a standard sphere and a plaster model with complex free-form surface are reported. The measurement results show that the measurement precision can achieve 0.095mm and the system can measure complex object in a short time with smooth details.Practical implications Complex free-form object high accurate 3D reconstruction in a short time with less sensitive to the non-sinusoidal of the fringe pattern is presented.Originality/value This work proposes a novel high accurate complex object 3D measurement method.

3D measurement Neural network Phase-height mapping Digital fringe projection

Z.W. Li D.H. Qin Y.S. Shi C.J. Wang G. Zhou K. Huang

State Key Laboratory of Material Processing and Die & Mould Technology,Huazhong University of Science and Technology, Wuhan, 430074, China.

国际会议

第三届北京国际快速成形及制造会议暨第二届北京国际生物制造会议(The 3rd International Conference on Rapid Prototyping and Manufacturing and The 2nd International Conference for Bio-manufacturing (ICRPM-BM 2008 Beijing))

北京

英文

166-171

2008-11-06(万方平台首次上网日期,不代表论文的发表时间)