Improving surface measurement by adaptive focal length adjustment and approximation networks

A three-dimensional model to perform measurements of variable range is presented. This model is performed by a Bezier network via projection of a laser line and an adaptive focal length. The variation of the measurement range is provided by the adjustment of the camera placement and the focal length. The modifications of these parameters are determined by image processing of the laser line during the vision task. The initial calibration of the vision parameters is deduced by means of the variation of focal length. The technique is applied to retrieve the contour of complex shapes by means of the adjustment of the measurement range. This technique avoids external re-calibration to compute the surface depth when the measurement range is adjusted. Therefore, the accuracy of the surface measurement is improved. It is because the errors of re-calibration are not added to the system during the measurement process. The change of the measurement range is achieved in fractions of a second. The contribution of this technique is stated by an evaluation based on the lighting methods that perform measurements of variable range. Also, the time to obtain measurements of variable range is described in the evaluation.

► We provide a modelling of variable measurement range based on variable focal length. ► The variable range is achieved by adjusting the camera position and the focal length. ► The model is performed by a Bezier network based on a laser line position. ► The focal length is online self-computed by image processing of a laser line. ► The proposed technique avoids re-calibration and improves the surface measurement.