An Efficient Method for Camera Calibration Using MultiLayer Perceptron Type Neural Network

Abstract

This paper presents a 3D camera calibration method based on a nonlinear modeling function of an artificial neural network. The neural network employed in this paper is primarily used as a nonlinear mapper between 2D image points and points of a certain space in 3D real world. The neural network model implicitly contains all the physical parameters, some of which are very difficult to be estimated in the conventional calibration methods. MutiLayer Perceptron Type Neural Network (MLPNN) is employed to implement the relationship between image coordinates. In order to show the performance of the proposed method, we carry out experiments on the estimation of 2D image coordinates given 3D real world coordinates. The experimental results show that the proposed method improved calibration accuracy over widely used Tsai's two stage method (TSM).

On the neural computation of the scale factor in perspective transformation camera model

“…Thus, we cannot train the network using any supervised learning technique such as a gradient descent algorithm. In [13], it is shown that a network can be trained if the z coordinate is set to a constant value. However, the network then can be applied only to some limited cases where the distance between a camera and target points is constant.…”

“…Although there are many camera models available, pinhole camera model is the most widely used mainly due to its simplicity [2,7,13,16].…”

“…To determine the elements of the camera matrix, the homogeneous image coordinates in (4) must be normalized, and then inhomogeneous image coordinates are obtained as  11 12 13 …”

“…In [13], ANNs are trained to learn the projection and back-projection of a camera. In [14,15], we have shown that an ANN can be successfully employed to calibrate stereo cameras.…”

The Development of Undistortion Digital Image Application using Backpropogation Neural Network Model

“…A solution to correct the geometrical distortion in image is typically used mathematical algorithm of polynomial equations to estimate and interpolate coordination of image [2][3][4][5]. An artificial neural network (ANN) is widely applied to solve various engineering problems such as pattern recognition, optimization and modeling [6,7] and also in computer vision [8,9]. The ANN is suitable to be employed for solving the problems that require complex models.…”