Noise-compliant tip-shape derivation

Wide use of the blind tip-derivation procedure has lead to the realization of the problem of noise in experimental image data. A modification to the algorithm by which the probe shape is extracted from image data is described. By consideration of the possible levels of noise in an image, a realistic estimate of the probe geometry can be derived from unfiltered experimental image data. The new procedure is demonstrated by its application to a test image that has been deliberately degraded through the addition of random noise. The results are compared with those from the current technique. The technique has significant application in metrology through the compensation of both lateral and vertical measurement errors.