Visualisation and Animation of Undiscovered Tomb Interior at the Prague Castle

There are many tombs of Czech kings and princes at the Prague Castle. Some remains of the Czech princes and their wives from the Přemysl epoch (9 century-1306 A.D.) were transported to a new cathedral of St.Vitus (1373) by the emperor and Czech king Charles IV (1316-1378) in the Middle Ages. The tombstones were made in the well-known stone workshop of the Parléř family. During war in the year 1757 on wish of Fredrick II., the king of Prussia, some tombs were partly damaged and opened. Majority of the tombs were investigated in the 19 and 20 centuries. The tombs of Bořivoj II (1100-1124) and Břetislav II (1092-1100) were restored in 2002. Mixed stones with lime were discovered under the upper part of the tomb; the space for the remains was been found under this layer and under the floor level. In the case of tomb Břetislav II, the prince, the last unknown opening has been stated in the 19 century (a small lead tablet with the date was found). In the case of Bořivoj II, the prince, the tomb has not been opened for more than 600 years (from the Charles IV time). It is a single undiscovered Přemysl ́s tomb at the Prague Castle. This is because of large two stones (2x3m) shield the space over remains in the ground. The decision on status quo has been made after discussions. The reconnaissance using endoscopes has only been allowed. The 3D visualisation of the Bořivoj II tomb interior is the theme of this paper. 1. IMAGE DATA SET In spring 2002, the sculptures on the tombs of Bořivoj II (1100-1124) and Břetislav II (1092-1100) were restored. Within this work, discovering of the space for the remains in both cases was planned. Mixed stones with lime were discovered under the sculptures; the lower part of the tomb located under the floor (cca 100x50x100cm) contains the remains of Czech princes. The decision on status quo has been made after discussions. The reconnaissance using endoscopes has only been allowed. Only the images from this device were used for possible creation of a 3D model. For a small camera input, the 2cm distance between two stones covering the tomb was prepared. Finally, two types of miniature cameras were used: the XL PRO VideoProbe (Everest-Vit) and the MTV-54C5P (Mintron) camera. In both cases, video sequences were obtained. From these video sequences singe images were generated with resolution 768x575 pixels. The VideoProbe XL PRO is the most advanced remote viewing tool. It is especially designed to be easy to use and still offer unmatched imaging capabilities with options to suit virtually any applications. With number of accessories and options available, the XL PRO is ready to be adapted to any inspections situations, for example the using in archaeological purposes. The video bore scope consists of hand-piece buttons, joystick and onboard file manager, high resolution camera for true colour images, external monitor and recording device with S-video inputs. The image sensor contains from 1/6 ́ ́ colour CCD camera with resolution of 480 HTV lines S-video (440000 pixels PAL). Exposure from interval 1/60 seconds up to 12 seconds can be used. The XL PRO supports the use of interchangeable optical tips – the 3.9mm, 5mm, 6mm and 8mm diameter probe. The second used camera was MTV-54C5P (Mintron Enterprise Co.,Ltd). This device uses interchangeable objectives (f=16,12,8,6,4,2.5mm). The objective with focal length 6mm was used for majority of acquired images. Camera is build in a tube; it contains 1/4-inch CCD Image Sensor (PAL, CCD total pixel 542 (H) x 586(V), electronic shutter,1/50~1/120,000 sec., scanning with 625 lines, 50 fields/sec ). The camera MTV-54C5P is not a complex viewing tool; in this case, a special small halogen light source must be added. Video data were stored on digital video recorder. Fig.1 MTV-54C5P (Mintron) CCD camera Fig.2: XL PRO VideoProbe (Everest-Vit) bore scope 2. IMAGE PRE-PROCESSING Both above mentioned cameras are non-metric. In this case, for using in photogrammetry, it is necessary to make a precise camera calibration. In special photogrammetric software, usually a special equation form for the image distortion is used. If you need only simple photogrammetric processing, or it is not possible to make a correct camera calibration, the images without big distortion are necessary to use. However, in simple photogrammetric software (such as Photomodeler for intersection method), the images with corrected image distortion it is possible to use. We have made only simple camera calibration to decrease image distortion. A special method based on image calibration and self-made software was designed. As a reference image, the calibration field composed from a set of regular small crosses on a paper sheet was prepared. Equations (1.1) are completed with correction part for expression of image distortion (1.1): ( ) b d d 0 x x R x x ′ = − ⋅ − (1.1) ( ) b d d 0 y y R y y ′ = − ⋅ − ( ) ( ) 2 4 6 d d d d 0 1 2 R R x , y , r x , y k r k r k r ′ ′ = = ⋅ + ⋅ + ⋅ ( ) ( ) ( ) 2 2 d d d 0 d 0 r x , y x x y y = − + − xd, yd are the coordinates with distortion, xb, yb are the coordinates without distortion, x0, y0 are the coordinates of principal point, k0, k1, k2 are the coefficients of radial distortion. The collinear transformation for plane calibration field is used. In the equation (1.2), there are: X,Y calibration field points coordinates, x,y image coordinates and a,..,h transformation coefficients. 1 ax by c X gx hy + + = + + 1 dx ey f Y gx hy + + = + + (1.2) The equations of collinear transformation (1.2) are completed with correction part for expression of image distortion from (1.1): ( ) 0 1 d d ax by c x R x x gx hy + + ′ − − = + + (1.3) ( ) 0 1 d d dx ey f y R y y gx hy + + ′ − − = + + These equations are used in the software for distortion parameters calculation. Fig.3 Simple radial image distortion suppressing (original image and processed image) For the image distortion parameters computation, the software K3 has been prepared. As a data input, the set of true coordinates from calibration field and the set of image coordinates of crosses in text format are used. The next step is the elimination of known radial distortion from images by using of another software. This software creates a new image without distortion. For every pixel of undistorted image, the identical point from distorted image is computed. This software uses inverse equation to equation (1.1): ( ) d b d 0 x x R x x ′ = + ⋅ − (1.4) ( ) d b d 0 y y R y y ′ = + ⋅ − xd, yd (the coordinates with distortion) are on both sides, the iteration is necessary to use : (1.5)