[vsnet-chat 3825] Blooming and Streaks Cancel Filter

[Seiichi Yoshida <comet@aerith.net>]

MISAO Project Announce Mail (November 28, 2000)

Hello. I am Seiichi Yoshida working on the MISAO project.

Two new image processing filters, blooming cancel filter and streak
cancel filter, are embedded to the PIXY System 2, the automated
image examination system.

When a very bright star comes in the field of the CCD image, it causes
blooming or streaks.

The blooming and streaks are not only eyesore to appreciate the image,
but also cause some problems while automated image examination using
the PIXY System.

The blooming is the intensive vertical line of the flood light from
the saturated bright star. Because the PIXY System measures the
position of a star as the gravity center, it is shifted along the
intensive line when the blooming appears. As a result, the system
cannot find the data in the star catalog at the measured position, and
outputs the star as a candidate of new object.

The streak is a belt of light spreading horizontally to one side from
a bright star, like a cometary tail, as being blurred by a
brush. Because the pixel value is larger than usual in the streak
areas, the system sometimes detects noises and outputs as candidates
of new objects by mistake.

The two filters embedded this time deletes the blooming and streaks
on the image automatically, without any parameter setting or try and
errors. Therefore, it is useful for automated examination of many
images as the MISAO Project does.

Some results of the filters are available at the following page.

  http://vsnet.aerith.net/misao/pixy/result/blooming.html

In the image of M45, many stars causes blooming. But they are deleted
by the blooming cancel filter.

In the image of 2000 UV13, both the blooming and streaks appear at the
same time. Operating the streak cancel filter at first, then operating
the blooming cancel filter, and most of the blooming and streaks are
deleted. In addition, the cross pattern by the sub mirror and the
stars in the streak area are remained properly after the operation.

In the image of M31, no streak appears and there are faint nebulae.
It shows that the streak cancel filter does not delete the nebulae.

# Many thanks to KenIchi Kadota and Toshikazu Muramatsu for the images
  used in that page.

All of the source files of the PIXY System are open in the public. So
the source files of the two filters are also available at the PIXY
System Home Page.

Unfortunately, operation of the two filters, especially of the streak
cancel filter, takes some minutes, about between one minute and ten
minutes. The time changes based on the computer.

In the course of the MISAO Project, new objects are searched only when
there are some past images of the same area in the database. But
actually, much more images are offered which has no past images of the
same area. We have discovered more than 1000 new variable stars so
far, but the examined images are only small part of the total offered
images.

In case there is no past images of the same area, only we can do is to
compare to the star catalog. However, because a red star appears
extremely brighter on the CCD images than photos, most of the stars
which are not recorded in star catalogs are not real new objects.
For example, we have found many 12 mag stars not recorded in the
USNO-A1.0 catalog, which covers down to 20 mag.

But for example, if a 6 mag star not recorded in star catalogs is
detected, it may be a remarkable new object like a nova because it is
too bright (Of course, checking minor planets and so on is required). 
So even if there is no past images of the same area, a very bright
star not recorded in star catalogs is worth being confirmed. But such
a bright star always causes blooming. The old PIXY System measured the
shifted position of such blooming stars, and all bright stars were
output as new objects. Therefore, real bright new objects have been
missed.

Now that the blooming cancel filter deletes the blooming on the image,
the shift of position will not happen. Then we can check at least the
existence of unknown bright objects in all of the offered images to
the MISAO Project.

By the way, the current version has some limitations.

- The streaks must spread horizontally.
- The blooming must stretch vertically.
- The blooming stretching through two or more bright stars, or the
  blooming without the star itself in the image field are not
  supported.

Because not too many images have been operated by these filters, I
think they cannot work well on some images. Then please make a contact
to Seiichi Yoshida.

P.S.
The past MISAO project announce mails are available at:
  http://vsnet.aerith.net/misao/

--
Seiichi Yoshida
comet@aerith.net
http://vsnet.aerith.net/

VSNET Home Page

Return to Daisaku Nogami

---

vsnet-adm@kusastro.kyoto-u.ac.jp