Fix sampling for version multiplying factor #22025

AleksandrPanov · 2022-05-23T17:35:30Z

Merge with extra: opencv/opencv_extra#976

Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

I agree to contribute to the project under Apache 2 License.
To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
The PR is proposed to the proper branch
There is a reference to the original bug report and related work
There is accuracy test, performance test and test data in opencv_extra repository, if applicable
Patch to opencv_extra has the same branch name.
The feature is well documented and sample code can be built with the project CMake

alalek · 2022-05-28T11:39:41Z

modules/objdetect/src/qrcode.cpp

+    for (int i = 0, r = 0; i < version_size; i++ && r < postIntermediate.rows, r+= delta_rows)
    {
-        for (int c = 0; c < postIntermediate.cols; c += delta_cols)
+        for (int j = 0, c = 0; j < version_size; j++ && c < postIntermediate.cols, c+= delta_cols)


Overcomplicated (need to rewrite) or just invalid "for" loops.

alalek · 2022-06-03T10:18:19Z

modules/objdetect/perf/perf_qrcode_pipeline.cpp

@@ -106,10 +106,11 @@ PERF_TEST_P_(Perf_Objdetect_QRCode_Multi, decodeMulti)
 INSTANTIATE_TEST_CASE_P(/*nothing*/, Perf_Objdetect_QRCode,
    ::testing::Values(
        "version_1_down.jpg", "version_1_left.jpg", "version_1_right.jpg", "version_1_up.jpg", "version_1_top.jpg",
-        "version_5_down.jpg", "version_5_left.jpg", "version_5_right.jpg", "version_5_up.jpg", "version_5_top.jpg",


We could use C++ comments here: /* "version_5_right.jpg", */ instead of whitespace

alalek · 2022-06-03T10:18:51Z

modules/objdetect/src/qrcode.cpp

@@ -2369,7 +2369,8 @@ bool QRDecode::samplingForVersion()
    CV_TRACE_FUNCTION();
    const double multiplyingFactor = (version < 3)  ? 1 :
                                     (version == 3) ? 1.5 :
-                                     version * (version + 1);
+                                     (version < 6) ? version*(version+1) :
+                                     version;


version;

version * 5 ?

fixing the tail processing process allows to reduce multiplyingFactor to 1 and skip strange resize() step

I want to add a few more tests and skip resize() step or reduce the multiplyingFactor to 2, for example.

now multiplyingFactor:

const double multiplyingFactor = (version < 3) ? 1. : (version == 3) ? 2. : 3.;

alalek · 2022-06-03T10:21:06Z

modules/objdetect/src/qrcode.cpp

+        deltas_rows[(i*version_size)/abs(skipped_rows)+(version_size/abs(skipped_cols)-1)/2]
+        += skipped_rows > 0 ? 1 : -1;


Please separate index calculation to simplify code.

for (int i = 0; i < abs(skipped_rows); i++) { // fix deltas_rows at each skip_step const double skip_step = static_cast<double>(version_size)/abs(skipped_rows); const int corrected_index = static_cast<int>(i*skip_step + skip_step/2); deltas_rows[corrected_index] += skipped_rows > 0 ? 1 : -1; }

alalek · 2022-06-03T10:26:52Z

modules/objdetect/src/qrcode.cpp

+        += skipped_cols > 0 ? 1 : -1;
+    }
+
+    const double totalFrequencyElem = countNonZero(postIntermediate) / static_cast<double>(postIntermediate.total());


Looks like postIntermediate is not a binary (0/255) image. So there is no sense to use countNonZero() here.

Perhaps we want to use norm(postIntermediate, NORM_L1 | NORM_RELATIVE); here.

Do we have 0.5 value here?

actually postIntermediate is binary (0/255) image in this case:

no_border_intermediate is binary

postIntermediate = resize(no_border_intermediate, newFactorSize, INTER_AREA)

newFactorSize is integer and InterpolationFlags is INTER_AREA, so postIntermediate is binary (0/255) image

I checked this in tests. Also totalFrequencyElem must be around 0.5, otherwise most of the tests will fail.

…ression_version_25

AleksandrPanov · 2022-06-03T16:48:09Z

@alalek, fixed the issues

AleksandrPanov force-pushed the fix_samplingForVersion_multiplyingFactor branch from 855f948 to 09e0eb7 Compare May 27, 2022 13:44

Kumataro mentioned this pull request May 28, 2022

objdetect: Fix sampling QR code image #21949

Closed

6 tasks

alalek reviewed May 28, 2022

View reviewed changes

AleksandrPanov force-pushed the fix_samplingForVersion_multiplyingFactor branch 3 times, most recently from 80e437f to 15b05c3 Compare June 1, 2022 23:47

reduce experimentalFrequencyElem and listFrequencyElem

9b02b93

AleksandrPanov force-pushed the fix_samplingForVersion_multiplyingFactor branch from 15b05c3 to 9b02b93 Compare June 2, 2022 02:11

AleksandrPanov added 2 commits June 2, 2022 05:21

fix large resize

051d34d

fix tile in postIntermediate

311a12b

AleksandrPanov added the test label Jun 2, 2022

alalek reviewed Jun 3, 2022

View reviewed changes

AleksandrPanov added 2 commits June 3, 2022 18:20

add getMinSideLen(), add corrected_index

73b38b7

add test decode_regression_21929 author Kumataro, add test decode_reg…

98a95d6

…ression_version_25

AleksandrPanov force-pushed the fix_samplingForVersion_multiplyingFactor branch from 27e28c2 to 98a95d6 Compare June 3, 2022 16:13

AleksandrPanov marked this pull request as ready for review June 3, 2022 16:44

alalek approved these changes Jun 4, 2022

View reviewed changes

alalek merged commit 53eda42 into opencv:3.4 Jun 4, 2022

alalek mentioned this pull request Jun 4, 2022

(4.x) Merge 3.4 #22067

Merged

This was referenced Jun 8, 2022

Swiss QR-bills (ISO 20022) not decoded #21929

Closed

Doesn't correctly recognize some QR #21417

Open

QRCodeDetector fails randomly (+samples) #18372

Open

Unable to get result from QRCodeDetector.detectAndDecode() when encoded QR Code data is long #17152

Closed

alalek mentioned this pull request Aug 21, 2022

(5.x) Merge 4.x #22408

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix sampling for version multiplying factor #22025

Fix sampling for version multiplying factor #22025

AleksandrPanov commented May 23, 2022 •

edited by alalek

Loading

alalek May 28, 2022

AleksandrPanov Jun 3, 2022

alalek Jun 3, 2022

AleksandrPanov Jun 3, 2022

alalek Jun 3, 2022

AleksandrPanov Jun 3, 2022 •

edited

Loading

AleksandrPanov Jun 3, 2022

alalek Jun 3, 2022

AleksandrPanov Jun 3, 2022

AleksandrPanov Jun 3, 2022

alalek Jun 3, 2022

AleksandrPanov Jun 3, 2022 •

edited

Loading

AleksandrPanov commented Jun 3, 2022

		deltas_rows[(i*version_size)/abs(skipped_rows)+(version_size/abs(skipped_cols)-1)/2]
		+= skipped_rows > 0 ? 1 : -1;

Fix sampling for version multiplying factor #22025

Fix sampling for version multiplying factor #22025

Conversation

AleksandrPanov commented May 23, 2022 • edited by alalek Loading

Pull Request Readiness Checklist

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

AleksandrPanov Jun 3, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

AleksandrPanov Jun 3, 2022 • edited Loading

Choose a reason for hiding this comment

AleksandrPanov commented Jun 3, 2022

AleksandrPanov commented May 23, 2022 •

edited by alalek

Loading

AleksandrPanov Jun 3, 2022 •

edited

Loading

AleksandrPanov Jun 3, 2022 •

edited

Loading