sikuli-driver team mailing list archive

[RaiMan <question228458@xxxxxxxxxxxxxxxxxxxxx>]

Question #228458 on Sikuli changed:
https://answers.launchpad.net/sikuli/+question/228458

    Status: Open => Answered

RaiMan proposed the following answer:
In Sikuli you have the following measures to improve robustness:
--1. only search in the region, where you expect your targets to appear
--2. use shots, that concentrate on the key visual aspects of the image - as little (unnecessary) background as possible
--3. work with similarities as high as possible (values > 0.9), to avoid false positives 

with your current challenge of same symbols with alternating background in a column measures 1 and 2 are mandatory.
The similarity should not be a problem (but might be) with an optimized screenshot. This is what you have to test and find out.

example:
- using a shot made on white background
- match similarity on white background stripes will be 1.0
- match similarity on purple background might be between 0.85 and 0.95 (e.g. 0.92)

- using a shot made on purple background the results will be vice-versa
and may be different.

So you will use:

matches = column_region.findAll(Pattern(shot).similar(0.9))

This should give you all symbol matches (which you can sort according to
the answer of j-the-k in your other question).

And in the loop, you can check the similarity and even know, wether it
is a white or a purple strie you are in currently.

-- 
You received this question notification because you are a member of
Sikuli Drivers, which is an answer contact for Sikuli.