[Toriyama (Aoyama Gakuin Univ.), Yuzurihara]
This is yesterday's report.
I tried to identify the location of mirror center on Tcam camera, by using the opencv python package.
Top figure shows the original figure.
Middle figure shows the binarized figure by applying gray scale. But the identification is not yet.
Bottom figure shows the another trial for the identification on PowerPoint.
Japanese explanation is attached.
[Toriyama, Yuzurihara]
Toriyama-san identified the center of ETMX mirror on Tcam figure, by adjusting the circle on Power Point. (see the top figure) Tomorrow, we will put the sticker on the k4mon display.
We are also tring to identify the mirror location by using opencv python package. The second image shows the mirror that was processed to remove fine light by gray processing, applying gaussian blur, and then binarizing.
The Japanese explanation is attached.
[Toriyama, Yokozawa]
Toriyama-san put the sticker on the k4mon display. (see the attached figure)
[Toriyama, Yuzurihara]
Toriyama-san worked on the automatic identification of mirror location by using the image recognion technique. But, he found it is not easy project. Note that we got many inputs from Dan Chen-san yesterday.
He changed the starting point from the Tcam figure to simulated figure which was made on power point. As shown in the third figure on attached pdf, he prepared the figure of the dottded circle.
He developed the python code for image recognion by using opencv.
1. The opencv identified the 21 edges of separated black islands.
2. The returned array has the positions of 21 edges. Becayse it is not easy to use, he converted the position information to barycenter information, which is shown in the first figure as black circle.
3. He fitted the these black circles by using least squares. The red large circle shows the fitted result. The center positions (x, y) and radius are obtained.
He also tried to apply same code to the image with only a part, which is shown in forth figure. It looks like working well ! But, the estimated center seems to be a bit away from the center estimated in the previous case. Next step is to apply the code to the full circle (with solid line) and to evaluate the quantitative error by comparing the cases of the solid line and image with only a part.
Note that he will stay Kamioka site until March 25.
[Toriyama, Yokozawa, Ushiba, DanChen, Oshino, Yuzurihara]
This is a report by Toriyama-san during last week. Sorry for late.
We are developping the program to find the edge of the mirror from a screenshot (before.png) of a monitor displaying ETMX in the control room.
(1) the center position of the circle in the screenshot is determined roughly by eye and cropped as after1.png.
(2) the figure is binarized. Figure (after3.png) shows the superposition of the original figure and the outline of each island.
(3) Find the average of the coordinates of each island contour and fit a circle by using the least squares method. Figure (estimated_circle.png) shows the result of fitting. The obtained circle and its center are displayed in the original image (result.png).
After that, we discussed the result with the commissioners (Ushiba-san and DanChen-san) and reaslized that the circle we estimated is actually the inner circle of the earthquake-stopper, not the mirror edge. We got the comments that we should not use the screenshot for this puepose, due to lack of the resolution. The fiure (ETMA_overvew.png) shows our understanding of ETMX.
We learned how to take picture of test mass by using Tcam on medm. We took pictures of ETMX and ITMX by changing the configuration (gain/exposure time/LED flash/green laser).
We again started to develop the code to find the actual mirror edge from the high resolution figure.
[Toriyama, Yokozawa, Ushiba, DanChen, Oshino, Yuzurihara]
Toriyama-san summarized about fitting the ITMX mirror to identify the center, as slide. (sorry for Japanese slide)
Requirement for Tcam ITMX is gain=470 and exposure time 20s.
By using his code, we can identify the center of ITMX mirror within the 1% error. The code is tested by the ITMX mirror which is intentionally moved by Ushiba-san.
