problem in setting 3 points about plotdigitizer

[dilawar]

Yes (0,0) is origin where both axes intersect.

For example, if we are going to click on points (0,0), (10,0) and (0,1). Use the following command.

./plotdigitizer.py -i image.png -p 0,0 -p 10,0 -p 0,1

If problem persists, please upload the original image as results (or send over via email). Thanks for your interest in this project; it will give me some motivation to polish it up.

[johnSmith1990]

thanks for your reply.
I am using your trimmed.png image.
my problem is that I dont know where do I click for those 3 points. when I run this command :
./plotdigitizer.py -i ./figures/trimmed.png -p 0,0 -p 10,0 -p 0,1
a window pop up that show image and wnat to click on(0,0)point. I click in the origin of graph. Is this correct?
then asks to click on (10,0)(second point). where do I click in the graph?
also for 3th point.

in the next command you mentioned that we can use this command if we know the locations of those points.
./plotdigitizer.py -i ./figures/trimmed.png -p 0,0 -p 10,0 -p 0,1 -l 10,10 -l 200,20 -l 10,27
How you get locations? 10,10 -l 200,20 -l 10,27
what is the differece between positions and locations?
excuse me

[dilawar]

I see. I am sorry I didn't read your issue carefully. Sorry for confusing documentation, I should have asked a friend to review it.

I am working on a simple GUI which should resolve this issue. Meanwhile, following should help with locations of points.

The red arrows tip are the points as mentioned in the arrow labels.

In the next command you mentioned that we can use this command if we know the locations of those points.

Sorry about this. By location, I mean the location of point in pixels. For example, if you open the above image in a image viewer and hover your mouse around (0,0) (point here), somewhere in your image-viewer, you can get the pixels values (location). These terminology are confusing. Here is a screenshot which explain it.

I open the image in GIMP and hover my mouse at point (0,0). At the left bottom, you will see location of this point as (149,366). By -l, I meant this location.

[johnSmith1990]

thanks a lot. first of all, I thanks to your patience.
I cloned this repository, and I want to repeat your example. I opened trimmed.png on paint(Win image editor) and get these locations (0,0) --> 22,294 , (10,0) --> 82, 294 , (0,1) --> 22,219 for those points.
Anyway, when I run this command
python plotdigitizer.py -i ./figures/trimmed.png -p 0,0 -p 10,0 -p 0,1 -l 10,10 -l 200,20 -l 10,27 on your trimmed.png image, I get this output : https://ibb.co/6wkVzrc

[johnSmith1990]

I think the locations of points are not correct because the (0,0) and (10,0) points must have same y value(axis) but it is 10 for (0,0) and 20 for (10,0). do you know what i want to say?

[dilawar]

Yes. I think I know. The geometrical axis and axis scheming used in various image viewing programs may not be same. For example, opencv (which I use in this app) uses different scheme: top-left rather than bottom left is (0,0) (see here https://stackoverflow.com/questions/25642532/opencv-pointx-y-represent-column-row-or-row-column) which may be different than paint. I can't be sure about this.

In next version, which will come with a simple GUI, I'll will remove the location altogether since it is confusing.

[johnSmith1990]

I think the locations of points are not correct because the (0,0) and (10,0) points must have same y value(axis) but it is 10 for (0,0) and 20 for (10,0). do you know what i want to say?

is this correct? it is other than coordinate difference. Actually I need script than Gui. bercause I have some other processing after that.
Thanks.

[dilawar]

Anyway, when I run this command
python plotdigitizer.py -i ./figures/trimmed.png -p 0,0 -p 10,0 -p 0,1 -l 10,10 -l 200,20 -l 10,27 on your trimmed.png image, I get this output : https://ibb.co/6wkVzrc

This is a bug. Fixing it right now.

[dilawar]

I've just pushed changes to master branch which should fix this issue. The README.md is clearer now but still needs some work.

Check the Makefile which has a test command which correct points and locations. Please reopen it again if there is any issue.

[johnSmith1990]

I am sorry,
my source image is : https://ibb.co/S0f20Fw
but i get this result : https://ibb.co/Lnn2mJc

what's the problem? Is it because of init point's selection?

[dilawar]

Thanks for the data. For images where there is sudden change, I need to implement a slightly different algorithm with different filtering parameters. Current version would not work efficiently on it. If it is not urgently required, can I postpone it to weekends? best, Dilawar

…

On Tue, Oct 1, 2019 at 7:13 PM johnSmith1990 ***@***.***> wrote: I am sorry, my source image is : https://ibb.co/S0f20Fw but i get this result : https://ibb.co/Lnn2mJc what's the problem? Is it because of init point's selection? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#1?email_source=notifications&email_token=AAG2VQPAFPQH53PVQRZ4EGDQMNHX7A5CNFSM4I4FKVQKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEABJ4BA#issuecomment-537042436>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAG2VQP7DCRN7OGAK6NXHVTQMNHX7ANCNFSM4I4FKVQA> .

-- Dilawar NCBS Bangalore

[johnSmith1990]

Sure,Thank you so much.
I need to study the code in detail.

[dilawar]

The color of trace is not uniform in un2.png file. I am adding a debug output below. The circles in the the plot shows the estimated points. As you can see there are quite a few missing values. If you have a thicker trace, it might work better.

That said, even this situation can be improved. I have to work on it a while.

[johnSmith1990]

thanks dear,
here is my original image: https://ibb.co/yncPwXh
I pre-process it. but about thicker trace: Do I binarizethat image so that convert grayscale to 0 and 255 intensities? Or dilate image?