Venus cloud tracking -- July 31, 2006

During the past few weeks, I have repeated the steps required to process images from the 7/13/04 observations and created additional vector fields of the cloud motion. I have attempted to use different images than before, and to redo the processing, alignment, and compositing completely from scratch to see what effect this may have on the resulting vector fields.

Contents:

1. Combining files
2. Selecting images
3. Aligning images
4. Making a flat and mask
5. Processing the bracket-gamma composite
6. Making composite images
7. Making movies
8. Enhancing composites
9. Making vector fields


1. Combining files

The first thing I did was to combine 4 already existing files of information about images into a single file. The files

time.txt : universal time of observation of each image,

1041 16:17:01.523765 977.02539608
1042 16:17:03.969390 977.06615650
1043 16:17:06.517683 977.10862805
1044 16:17:08.984247 977.14973745
1045 16:17:11.548302 977.19247170
etc...

coords.txt : center, radius, and error of circle fitted to bright limb of each image,

1041     316.51     220.44     167.81    0.24008
1042     317.38     220.55     167.34    1.36973
1043     317.21     220.64     167.61   0.270484
1044     317.15     221.30     167.03   0.262651
1045     317.79     221.41     167.20   0.255174
etc...

sharpness11.txt : estimated sharpness of each image,

1041 0.0275021
1042 0.0274982
1043 0.0272446
1044 0.0272411
1045 0.0277015
etc...

and bad.txt : list of unusable images,

1107
1110
1111
1112
1113
etc...

were merged into a single file called image_data.txt : information required for subsequent processing and alignment.

1041  16:17:01.523765     0.00000000   -61.51   -34.56   999.00   999.00  0.0275021000
1042  16:17:03.969390     0.04076042   -62.38   -34.45   999.00   999.00  0.0274982000
1043  16:17:06.517683     0.08323197   -62.21   -34.36   999.00   999.00  0.0272446000
1044  16:17:08.984247     0.12434137   -62.15   -33.70   999.00   999.00  0.0272411000
1045  16:17:11.548302     0.16707562   -62.79   -33.59   999.00   999.00  0.0277015000
etc...

In this new file, the first column is the image number, the second column is the UT, the third column is the UT converted to a decimal minute elapsed from the time of the first image, the fourth and fifth columns are the fitted bright limb coordinates converted to a shift required to center the inverted image, the sixth and seventh columns are the actual shifts required to center the flipped image as determined by FITSRegister, when available, or the value 999 otherwise, and the eighth column is the estimated sharpness.

Out of 2400 total images, the image_data.txt file contains entries for 2220. Many images were rejected automatically from their limb coordinates, either because the image was out of the frame, or because the coadds were not aligned (i.e. the telescope was moving). Some additional images were rejected because while the bright limb coordinates indicated the planet was in the frame, the dark limb was actually partly out of the frame.


2. Selecting images

Plotting sharpness

12 different methods were used to attempt to automatically estimate the sharpness of each candidate image. The one determined to be the 'best' used the fraction of energy of the image in the spatial frequency band 9-16 Hz, as determined from the squared magnitudes of the Fourier transform. This result appears to be the most consistent with rankng the images manually by eye.

  sharpness    shift method - before brgm
  sharpness2   shift method - after brgm - r=100
  sharpness3   shift method - r=75 
* sharpness4   stddevs of disks
* sharpness5   stddevs of more disks
  sharpness6   unsharp mask
  sharpness7   contrast
  sharpness8   entropy
* sharpness9   fft energy - 17-32 Hz
  sharpness10  fft energy - 33-64 Hz
* sharpness11  fft energy - 9-16 Hz
  sharpness12  fft energy - 5-8 Hz


(Click for full-size)

Selecting intervals

Here is a plot showing the 25 sharpest and 25 least sharp images in each of 8 time intervals throughout the night of 7/13/04. The duration of the observation was 133 minutes, so that each interval is about 17 minutes long. In some intervals, the sharpest images are all at about the same time, while in others they are spread out over the entire interval. There are also several irregular gaps between clusters of sharp images. This makes it difficult to choose a set of, say, 10 images of superior sharpness all at about the same time at regular intervals during the observation.


(Click for full-size)

For a previous analysis, it was decided to select a few groups of nearby sharp images from obvious peaks in the sharpness plot, even though these sets did not come at evenly spaced time intervals.


(Click for full-size)


(Click for full-size)

For the current analysis, it was decided to divide the observation into 20 minute intervals, and to automatically select the 10 sharpest images from the central 5 minute period of each interval. In this way the resulting composite images would be at regular time intervals, and would, with the exception of the last interval, be at about the same level of sharpness.


(Click for full-size)

The 70 images chosen were:

Interval 1:  1230 1231 1232 1227 1228 1229 1233 1221 1235 1225
Interval 2:  1763 1757 1759 1785 1789 1750 1756 1748 1765 1760
Interval 3:  2124 2123 2113 2140 2118 2121 2117 2112 2139 2116
Interval 4:  2492 2517 2519 2507 2525 2555 2518 2484 2493 2515
Interval 5:  2897 2896 2893 2869 2898 2887 2843 2892 2895 2868
Interval 6:  3205 3160 3212 3161 3204 3162 3206 3215 3159 3182
Interval 7:  3465 3470 3482 3481 3480 3466 3467 3478 3483 3471

Several of the images originally chosen for interval 1 had to be manually rejected because the edge of the dark limb was out of the frame. They were replaced by the next sharpest image in the list for that time interval.


Next page ->

İSky Coyote 2006