|
1.INTRODUCTIONIn order to carry out high-precision measurements with the aid of optical image processing, it is of decisive importance that the quality of the image to be measured is of particularly high quality. Low noise, contrast and other factors play a role. A major problem here are distortions in the image caused by the lens. Figure 1 shows an extreme case of distortion of a lens. In such a case, metric measurements in the image range from poor accuracy to useless measurement results. To remove the distortion, the camera system must be calibrated. This procedure requires a calibration target (Fig. 2). 2.USEFUL STRUCTURES FOR CALIBRATION TARGETSIn order to determine the exact distortion of a lens, it is necessary to determine the deviation from the actual target position for each location in the image. This requires calibration targets that fill the entire usable image area. Furthermore, the exact coordinates of all geometric elements in the calibration target must be known. Meaningful structures can be seen in the following images: 3.DETERMINATION OF DISTORTIONThe positions of the geometric elements are determined in the image by means of image processing algorithms. These algorithms indicate the deviations in the image from the target positions. The determined positions serve as support points to create a flat 2D distortion correction. Figure 9 shows the image acquisition of a point structure and Figure 10 shows the probed point elements through image processing. 4.QUALITY OF THE CALIBRATION TARGETSThe exact determination of the position of the geometric elements is of essential importance. Factors influencing the accuracy of the position determination include the image parameters: Sharpness and contrast in the image, artifacts in the image and other parameters. But also the quality of the individual geometry elements on the calibration targets is of great importance and the material on which the calibration structures are applied. The following pictures show example printouts on normal copy paper, on a foil and on photo paper. Printing was done with the inkjet printer HP K7100 and the laser printer Lexmark Optra T610(MS). With the UNI-VIS 250 coordinate measuring machine from Mahr OKM GmbH, the point elements of the printed calibration targets were precisely measured. This provides information on how precisely the position of the geometry elements could be printed with the printer. These values serve as reference values for quality assessment. Figures 14 - 16 show an impression of the image quality of the image recordings of the calibration structures. Since the material printed on (paper, foil), is not ideal to achieve the highest image quality, there are losses in the achievable measurement accuracy. Depending on the selected print resolution and the resolution of the optical system, dots of the printing process are more or less recognizable (Fig. 17). In comparison, a professionally manufactured calibration standard looks like Figure 18. The position deviations of the point elements, measured with the UNI-VIS 250 coordinate measuring machine from Mahr OKM GmbH, can be seen in image diagrams 19 - 23. 5.CONCLUSIONSIt is possible to produce calibration targets with commercially available equipment (printers, print media). Depending on the printing technology and printing medium, the position deviations of the geometric elements can be an order of magnitude worse than those of a professionally produced calibration standard. Depending on the printer, periodic deviation errors can occur, which exist as systematic errors in addition to random errors. Thus, this represents a cost-effective and practicable solution for the simple production and use of calibration targets. REFERENCESSteger, Carsten, Ulrich, Markus, Wiedemann, Christian, Wiley-VCH Verlag GmbH & Co. KgaA, Weinheim
(2007). Google Scholar
Jähne, Bernd, Springer-Verlag, Berlin
(2012). Google Scholar
Zhang, Z,
“A flexible new technique for camera calibration. IEEE Trans. On Pattern Analysis an Machine Intelligence (PAMI),”
22
(11), 1330
–1334
(2000). Google Scholar
Tu, Huanbo,
(2014). Google Scholar
Steinbrecher, Rainer, Oldenbourg Verlag, München
(1993). Google Scholar
|