Profilometry for the measurement of three-dimensional object shape using radial basis function, and multi-layer perceptron neural networks

Phase-height relationship by plane analysis in 3D shape measurement using fringe pattern projector

“…Another possibility is the use of an artificial neural network to estimate the depth maps from the recovered phase distributions. 19 In this case, however, a timeconsuming training procedure must be performed.…”

“…If the fringe pattern of the projector is set to be parallel to the v-axis of the camera, all coefficients of Eqs. (19) and (20) are the same regardless of the v-axis Generally, it is possible to set some of the fringe patterns to be parallel to the v-axis of image plane, but it is nearly impossible to set all the patterns due to lens distortion. When the phase values of the fringe pattern are not constant along the v-axis, we must find modeling equations for the sampled v values.…”

Complex object 3D measurement based on phase-shifting and a neural network

“…As a neural network has merits of high-speed parallel calculative activity, powerful ability of function approximation and noise resisting [10], it has been introduced to solve the complex nonlinear phase-height mapping issue [11][12][13][14][15]. Refs.…”

“…Refs. [12][13][14] establish the mapping relationship between phase map or fringe image and height distribution using a multi-layer neural network, thus obtaining the height of the object directly without knowing the system parameters. But they are time-consuming and it is hard to collect good sample data for neural network training.…”

BP neural network applied to 3D object measurement based on fringe pattern projection

“…Using this method, the depth of the object can be recovered by integrating the height directional gradients obtained from network. Ganotra et al [7] used RBF neural network and multi-layer neural network to measure the height of the object. At first, they use the laser as light source to measure the height of object.…”