MISAO Project

Home Page       Sat Dec 6 19:31:03 JST 1997

next up previous
Next: Matching of peculiar images Up: Interim Report on Matching Previous: Overview of matching

Experimental results of matching


I made experiments with these images, if they can be matched properly with star charts. Please see the pictures page of each object for detail data of each image.

Because these images were taken in order to catch known comets or novae, I input the ephemeris of each object on each day to the system as the approximate position, though the object is not absolutely in the center. The view width is different one another, but the photographers know the rough values and I input them to the system as the approximate view width.

These images were matched with the Tycho Catalog, except for b2_0327.jpg and 1996B2.jpg, matched with the Yale Catalog of Bright Stars, because they are extremely wide. As a result, the system succeeded to match properly almost all images, except for two images, and return the precise position, precise view width and position angle.

Here is an example. The image of Comet Tabur on Sept. 16, 1996, (tb960916.jpg) was matched with the Tycho Catalog. I input R.A.= 5h40m, Decl.=+7o for the approximate position, 5 deg for the approximate view width. The system succeeded to match and the result images is below, which is compressed half both vertically and horizontally. The left image is the original. The right one is the chart of the same area mapped with the position, view width and position angle the system returned after matching.

The coordinates of the center the system returned is: R.A.= 5h42m1.61s, Decl.=+6o38'1.4''. This image is arranged so that the view width should be 5 deg and north is up. However, as a result of this experiment, it was cleared up that the precise view width is 4.18 deg and the position angle of up direction on the image is 1.24 deg, both are slightly different.

Because this image has so many stars, it is very hard to determine which star on the chart is the counterpart of each star on the image without software support. When the image rotates or the view width is uncertain, matching becomes almost impossible. Therefore it is a great merit to match automatically with a software.

By the way the experiment was run with a PentiumPro 180MHz PC. Although the PIXY system is implemented in Java language, only the matching part is implemented in C language as a native method. Matching took some seconds or several minutes.

next up previous
Next: Matching of peculiar images Up: Interim Report on Matching Previous: Overview of matching

Copyright(C) Seiichi Yoshida (comet@aerith.net). All rights reserved.