Fix Hardsubx OCR

[hrideshmg]

In raising this pull request, I confirm the following (please check boxes):

• I have read and understood the contributors guide.
• I have checked that another pull request for this purpose does not exist.
• I have considered, and confirmed that this submission will be valuable to others.
• I accept that this submission may not be used, and the pull request closed at the will of the maintainer.
• I give this submission freely, and claim no ownership to its content.
• I have mentioned this change in the changelog.

My familiarity with the project is as follows (check one):

• I have never used CCExtractor.
• I have used CCExtractor just a couple of times.
• I absolutely love CCExtractor, but have not contributed previously.
• I am an active contributor to CCExtractor.

---

closes #1723

Hardsubx is currently broken on the master branch on Rust.

Fixed a segmentation faults on Linux which was caused by a null pointer dereference in dispatch_classifier_function, I've added proper error handling to fix this. While here, I also noticed that the values in the match statement for the classifier functions are incorrect, see here. The pixDilateGray() function also had an incorrect argument, see here.

While CCextractor seemed to run after this, the output seemed to be garbled. On running some tests, I found out that the luminance mask was fully white. This is because the Srgb::new() function expects values in the range of 0.0-1.0, however we were passing values from 0.0-255.0.

I've also enabled hardsubx for the test builds. This is to facilitate adding regression tests on the sample platform.

[ccextractor-bot]

CCExtractor CI platform finished running the test files on linux. Below is a summary of the test results, when compared to test for commit 4b5f68a...:

Report Name	Tests Passed
Broken	13/13
CEA-708	14/14
DVB	7/7
DVD	3/3
DVR-MS	2/2
General	27/27
Hauppage	3/3
MP4	3/3
NoCC	10/10
Options	86/86
Teletext	21/21
WTV	13/13
XDS	34/34

All tests passing on the master branch were passed completely.

Check the result page for more info.

[ccextractor-bot]

CCExtractor CI platform finished running the test files on windows. Below is a summary of the test results, when compared to test for commit 4b5f68a...:

Report Name	Tests Passed
Broken	13/13
CEA-708	14/14
DVB	4/7
DVD	3/3
DVR-MS	2/2
General	27/27
Hauppage	3/3
MP4	3/3
NoCC	10/10
Options	86/86
Teletext	21/21
WTV	13/13
XDS	34/34

NOTE: The following tests have been failing on the master branch as well as the PR:

• ccextractor --stdout --quiet --no-fontcolor 79a51f3500..., Last passed:

  Never

• ccextractor --stdout --quiet --no-fontcolor 767b546f96..., Last passed:

  Never

• ccextractor --autoprogram --out=srt --latin1 --quant 0 85271be4d2..., Last passed:

  Never

Congratulations: Merging this PR would fix the following tests:

• ccextractor --autoprogram --out=srt --latin1 f1422b8bfe..., Last passed: Never
• ccextractor --datapid 5603 --autoprogram --out=srt --latin1 --teletext 85c7fc1ad7..., Last passed: Never
• ccextractor --out=spupng c83f765c66..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 c0d2fba8c0..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 006fdc391a..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 e92a1d4d2a..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 7e4ebf7fd7..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 9256a60e4b..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 27d7a43dd6..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 297a44921a..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 efbe129086..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 eae0077731..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 e2e2b501e0..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 c6407fb294..., Last passed: Never
• ccextractor --autoprogram --out=ttxt --latin1 --datets dcada745de..., Last passed: Never
• ccextractor --autoprogram --out=srt --latin1 --tpage 398 5d5838bde9..., Last passed: Never
• ccextractor --autoprogram --out=srt --latin1 --teletext --tpage 398 3b276ad8bf..., Last passed: Never
• ccextractor --out=srt --latin1 f23a544ba8..., Last passed: Never
• ccextractor --out=srt --latin1 10f0f77cf4..., Last passed: Never
• ccextractor --out=srt --latin1 df3b4d62d3..., Last passed: Never

All tests passing on the master branch were passed completely.

Check the result page for more info.

[prateekmedia]

LGTM!

[rboy1]

Since this doesn't appear to be an exact port from C to RUST, some of the thresholds have changed and consequently the accuracy of the OCR has changed. i.e. you will see different results on C vs OCR builds.

So here my question, how does one tune the OCR thresholds? If the C version was working better than the RUST port there should be some way to tweak the thresholds.

[cfsmp3]

Since this doesn't appear to be an exact port from C to RUST, some of the thresholds have changed and consequently the accuracy of the OCR has changed. i.e. you will see different results on C vs OCR builds.

So here my question, how does one tune the OCR thresholds? If the C version was working better than the RUST port there should be some way to tweak the thresholds.

If the C version was working better then the Rust version is not ready, that's all.
This applies to all subsystems, of course.

Feel free to share samples and how to compare both versions, that's the best way to help getting us where we want to be.

[hrideshmg]

@rboy1 #1746 should fix the discrepancies between the Rust and C implementations. I'd also suggest playing around with --whiteness-thresh as well, setting it to lower values on certain files seemed to produce really good results.