Optical correlation-based measurement techniques by means of random dot pattern

The fundamental idea of the optical correlation-based measurement technique is to solve the correspondence problem by means of local cross-correlation. Therefore a random dot pattern is imaged under two different conditions using a digital camera. Besides the geometric calibration of the measurement system, the accuracy of the optical correlationbased measurement technique depends strongly on the accuracy of the cross-correlation algorithms, which provide accuracy In the sub-pixel domain. Therefore, the main focus within this work is on the investigation of the algorithms used for sub-pixel displacement estimation. It will be shown that previous algorithms for sub-pixel displacement estimation will be significantly biased in case of elliptically shaped correlation peaks. Additionally, an analytical derivation of the bias error is also included in this work. The analytical description of the bias error will be investigated by means of a numerical simulation.