[request] want a Function visible() like similar() to differentiate partially transparent from opaque images

[RaiMan]

Hai Raiman, Now a days all the websites having the css and JavaScript enabled , so that the visibility is adjusted as per their requirement. While we using Sikuli to search a image which is full visible other than partial visible image look like same. Here the similar function capture the partially visible also as it looks like the same image we search.If we adjust the similarity to low it allow non similar image also. Do you understand what I am Asking? Can you add this visible() feature. I know little python if you want my help I am eagerly waiting. I thinking this can be possible ? Is it possible in your Point of View?

[RaiMan (raimund-hocke)]

Not fully understood, what you are talking about.

Do you have some example or screenshot (can be added here)?

[sakthivel (sakthi-saran65)]

Raiman , Image in the attachment is example situation. That image is a screen shot of a web page. where there is all square in green color. But three of the square in less green (i.e 0.4 Opacity). I want to avoid that three squares and get all other squares in the findAll() function.

But what happening is all squares got into the findAll() even similarity set to 0.99. the similar() only considering shape and color.
It does not consider the how much Opacity it has. I suggest a visible() function to find the visibility that it is visible with full opacity or less opacity. So that we can filter the low visibilty squares from all others.

If this feature can be added. it should be place in near to similar. so that we can set visibility of the image for matching.

[RaiMan (raimund-hocke)]

I already was afraid, that you meant opacity/transparency.

Currently, the OpenCV feature used completely ignores the transparency channel in an image to be searched. Only the RGB part of an RGBA image is taken and hence as if it where 100% opaque (0% transparency).

An Implementation (if any) is very complex and might not be available before end of this year.

[sakthivel (sakthi-saran65)]

Thanks Raiman. I am waiting for version SikuliX 1.2.0. And also I have many ideas but I don't have enough programming skills to make it possible. if you don't mind I will post my ideas what I feel most important for Sikulix.

[RaiMan (raimund-hocke)]

@sakthi
Your are welcome, to post any request for features:
- use [request] as start for the title
- only one feature or feature group per request

BTW: the one or other feature might already be available in Beta versions earlier. So watch out ;-)

[sakthivel (sakthi-saran65)]

Ya Raiman now I am using the Beta version only. And I am checking every functions with some fixed target motive. In all test SikuliX doing very very good. Once again Thanks to You and SikuliX Drivers Team for this awesome job. In future I would like to contribute More.

[RaiMan (raimund-hocke)]

--- And I am checking every functions with some fixed target motive.
I did not have the time yet, to implement any tests, to e.g. avoid regressions.
So I am interested in what you are doing here.
You might contact me directly using my mail at https://launchpad.net/~raimund-hocke

--- the next days I will publish nightly builds somewhere in the net, so interested people might check fixes immediately without the need to build from sources.