|Search for content and authors
Sub-pixel detection of a grid’s node positions for optical diagnostics
|István E. Lukács , Ferenc Riesz
Hungarian Academy of Sciences, Research Institute for Technical Physics and Materials Science, P.O.Box 49, Budapest H-1525, Hungary
Determination of the position of a square grid’s nodes is a frequently encountered image processing task in optical metrology and diagnostics. For example, many optical shape metrology tools use a grid projected onto the object; further, grids are used in instrumentation and processes for alignment purposes. The accuracy of the grid detection greatly influences the accuracy of the whole measurement system and the speed of the grid detection determines the throughput of the measurement or process.
In the present paper, we describe a complete grid detection algorithm. Our research is motivated by application in quantitative Makyoh topography ; this method uses the projected image of a grid to determine the height topography of closely planar mirror-like surfaces. First, the software generates the correlation function using a special (cross-like) weight function, then, it finds the local maxima of the correlation function with one-pixel accuracy. Then, the neighbourhood relations are determined for the proper ordering of the data into a matrix. Finally, the algorithm refines the positions of the detected grid points with sub-pixel accuracy.
We have developed several versions of the algorithm for different tasks. We study and compare the accuracy and speed of the algorithms as a function of grid geometry, grid distortion, grid defocus, image noise and contrast using simulated images. Experimental application in quantitative Makyoh topography using semiconductor wafers are demonstrated as well.
1. F. Riesz, Proc. SPIE 5458 (2004) 86.
Presentation: Poster at E-MRS Fall Meeting 2007, Symposium H, by István E. Lukács
See On-line Journal of E-MRS Fall Meeting 2007
Submitted: 2007-05-11 14:48 Revised: 2009-06-07 00:44