Correction of projector’s nonlinearities in the vision based system for positioning of the workpieces on the CNC machines

In several last years a huge expansion of the optical measuring systems in industry and manufacturing is observed. The most advanced systems are the 3D geometry scanning systems based on images captured by cameras. A special group of them use structural light patterns projected onto the scanning surface and observed by the cameras. An appropriate coding and analysis of images allows obtaining any information about shape and size of the scanned object in relatively short period of time. Nevertheless, the 3D scanning accuracy strongly depends on complicated calibration process of the whole system, including a precise calculation of intrinsic and extrinsic parameters and correction of lens distortions both for camera and projector. The structural patterns emitted from the projector include gradients in a specified scale. Thus, using proper algorithms each pixel can be indexed and the precise depth map can be obtained. Using sinusoidal fringes illuminating the scanning surface, performing the linearisation of projector’s response is necessary in order to minimize the distortions of the sinusoidal waves. The correction method turns out to be difficult since the process depends on surface fracture, colour, and reflectivity. In this paper three methods for correction of the projector’s nonlinearity will be presented. The first one assumes a model of the nonlinear characteristics created from the gradient projected on a flat surface. The second approach is based on a series of captured images of scanned object illuminated by linearly increasing light allowing creating an individual correction curve for each pixel. The last method assumes an ideal inverted gamma correction model modified by dynamically changing coefficients calculated for each pixel. The efficiency and limitations of each method depending on the configuration of the system are also discussed in the paper as well as some experimental results obtained in the experimental vision based system for positioning the workpieces on the CNC machines.