Merge pull request #29335 from amd:fast_norm_simd

core: SIMD optimizations for norm, distance and Hamming APIs. (Improves ORB & BRISK) #29335

Generic universal-intrinsic kernels (benefit all SIMD backends: NEON, AVX2, AVX-512, etc.), plus enabling wider dispatch for the norm module.

- hal::normHamming: cached-pointer dispatch (resolve once, no per-call dispatch chain or trace region) + vector popcount path. cv::norm(NORM_HAMMING) and binary-descriptor matching (BFMatcher ORB/BRISK/FREAK via cv::batchDistance).
- hal::normL2Sqr_ / normL1_: direct inlinable kernels with single-vector tail (cv::batchDistance / BFMatcher float L2/L1, cv::kmeans).
- cv::norm masked NORM_INF: deinterleave SIMD for multichannel + back-step tail.
- cv::norm(src1, src2, type, mask): SIMD masked norm-diff; INF is one templated kernel for all element types, plus uchar L1/L2 and int L1 kernels.
- Enable AVX512_SKX/AVX512_ICL dispatch for the norm module.
- features2d: add BFMatcher knnMatch perf tests (float L2/L1, binary Hamming).
- ORB and BRISK performance improved
### Pull Request Readiness Checklist

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

- [x] I agree to contribute to the project under Apache 2 License.
- [x] 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
- [x] The PR is proposed to the proper branch
- [ ] There is a reference to the original bug report and related work
- [x] There is accuracy test, performance test and test data in opencv_extra repository, if applicable
      Patch to opencv_extra has the same branch name.
- [x] The feature is well documented and sample code can be built with the project CMake

modules/features2d/perf/perf_matcher.cpp

@@ -0,0 +1,91 @@
+// This file is part of OpenCV project.
+// It is subject to the license terms in the LICENSE file found in the top-level directory
+// of this distribution and at http://opencv.org/license.html.
+//
+// Copyright (C) 2026, Advanced Micro Devices, Inc., all rights reserved.
+
+// Brute-force descriptor matching perf tests. These exercise the core distance
+// kernels (hal::normL2Sqr_ / normL1_ / normHamming via cv::batchDistance) that
+// back BFMatcher, which is how float descriptors (SIFT 128-d, SURF 64-d) and
+// binary descriptors (ORB 32-byte) are matched.
+
+#include "perf_precomp.hpp"
+
+namespace opencv_test
+{
+using namespace perf;
+
+// (descriptor dimension, query/train descriptor count)
+typedef tuple<int, int> Dim_Count_t;
+typedef TestBaseWithParam<Dim_Count_t> DescriptorMatcherFixture;
+
+// Float descriptors matched with L2 (SIFT=128, SURF=64) — uses hal::normL2Sqr_.
+PERF_TEST_P(DescriptorMatcherFixture, bfmatch_knn_L2_float,
+            testing::Combine(
+                testing::Values(64, 128),       // SURF, SIFT descriptor sizes
+                testing::Values(512, 1000)
+            ))
+{
+    const int dim   = get<0>(GetParam());
+    const int count = get<1>(GetParam());
+
+    Mat query(count, dim, CV_32F);
+    Mat train(count, dim, CV_32F);
+    declare.in(query, train, WARMUP_RNG);
+    declare.time(60);
+
+    BFMatcher matcher(NORM_L2, false);
+    std::vector<std::vector<DMatch> > matches;
+
+    TEST_CYCLE() matcher.knnMatch(query, train, matches, 2);
+
+    SANITY_CHECK_NOTHING();
+}
+
+// Float descriptors matched with L1 — uses hal::normL1_.
+PERF_TEST_P(DescriptorMatcherFixture, bfmatch_knn_L1_float,
+            testing::Combine(
+                testing::Values(64, 128),
+                testing::Values(512, 1000)
+            ))
+{
+    const int dim   = get<0>(GetParam());
+    const int count = get<1>(GetParam());
+
+    Mat query(count, dim, CV_32F);
+    Mat train(count, dim, CV_32F);
+    declare.in(query, train, WARMUP_RNG);
+    declare.time(60);
+
+    BFMatcher matcher(NORM_L1, false);
+    std::vector<std::vector<DMatch> > matches;
+
+    TEST_CYCLE() matcher.knnMatch(query, train, matches, 2);
+
+    SANITY_CHECK_NOTHING();
+}
+
+// Binary descriptors matched with Hamming (ORB/BRISK=32 bytes) — uses hal::normHamming.
+PERF_TEST_P(DescriptorMatcherFixture, bfmatch_knn_Hamming_binary,
+            testing::Combine(
+                testing::Values(32, 64),        // ORB (32), BRISK/FREAK (64) byte sizes
+                testing::Values(512, 1000)
+            ))
+{
+    const int bytes = get<0>(GetParam());
+    const int count = get<1>(GetParam());
+
+    Mat query(count, bytes, CV_8U);
+    Mat train(count, bytes, CV_8U);
+    declare.in(query, train, WARMUP_RNG);
+    declare.time(60);
+
+    BFMatcher matcher(NORM_HAMMING, false);
+    std::vector<std::vector<DMatch> > matches;
+
+    TEST_CYCLE() matcher.knnMatch(query, train, matches, 2);
+
+    SANITY_CHECK_NOTHING();
+}
+
+} // namespace