MISAO Project

Home Page       Fri May 22 10:10:11 JST 1998

next up previous
Next: Conclusion Up: Database Astronomy using HORB Previous: Importance of network

PIXY system

Figure 1: Star catalog for the Hubble Space Telescope plotted on an image

Here is the overview of the system. Definitely, with a file of an image and some additional information, approximate position of the image in the sky, etc., it compares it with a star catalog and replies the data, position and brightness, of every star in the image, and a list of candidates of new stars which are not in the catalog. It also enables to read various kinds of data from various WWW sites and plot them on the image (Figure 1). Then the user can see the information of each star in the image by clicking a mouse on it.

The process consists of three steps. First, the system detects all stars from an image. Then it matches the detected stars and records in a star catalog and finds out the accurate position of the image in the sky. Finally it measures position and brightness of every star detected from the image. As a result, almost all detected stars have their own counterparts in the catalog, and those which are only in the image or only in the catalog are remarkable.

The system is in Java. Based on the object-oriented style of Java, the total system is assembled as a set of some modules. Each module receives an output of a module, does some process according to its own function, then gives the result to another module. Changing the user interface becomes easy in this way. Actually, both a usual Java application version and an applet version are available of this system.

The modules structure and data flow is at Figure 2.

Figure 2: Modules structure

Actually the star detection process and the matching process take much time. Implementing those modules in Java, the system took several hours to finish on PentiumPro 180MHz PC. Therefore the two modules are in C and called as native methods from the total system.

Because applets cannot read data from any WWW sites, and two main modules are implemented as native methods, and of course, in order to gather all data in the public, the system is in server/client model. Figure 3 shows the model of the applet version. HORB is used for communication between the server and the client.

Figure 3: Server/Client model

In this case, the image and the logs, midway progress of server's process and the result lists, are transferred from server to client. The size of an image is usually several dozens of kilobytes in GIF or JPEG format, or about 1 MB in FITS format, a standard in astronomy. The size of logs is about 100 KB usually. They are too large to transfer over the current Internet. Even in the case of non-applet version, the system has to send the results to the server for gathering all data, so the same problem happens.

next up previous
Next: Conclusion Up: Database Astronomy using HORB Previous: Importance of network

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