Merge pull request #3983 from mitresthen:warn-warpaffine-overlap

Add assert to ensure using non-overlapping memory regions #3983

This pr addresses this issue: opencv/opencv#27429
where the user did not realize that the warpaffine function requires non-overlapping src and dst memory regions. 
The code now compares the input memory regions and asserts that they do not overlap. There is also a test for this functionality.

### 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
- [x] 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.
- [ ] The feature is well documented and sample code can be built with the project CMake

Author: mitresthen

modules/cudawarping/include/opencv2/cudawarping.hpp

@@ -118,6 +118,7 @@ CV_EXPORTS_W void resize(InputArray src, OutputArray dst, Size dsize, double fx=
 @param src Source image. CV_8U , CV_16U , CV_32S , or CV_32F depth and 1, 3, or 4 channels are
 supported.
 @param dst Destination image with the same type as src . The size is dsize .
+    **In-place operation (src == dst) is not supported and will result in an error.**
 @param M *2x3* Mat or UMat transformation matrix.
 @param dsize Size of the destination image.
 @param flags Combination of interpolation methods (see resize) and the optional flag
@@ -127,6 +128,7 @@ INTER_NEAREST , INTER_LINEAR , and INTER_CUBIC interpolation methods are support
 @param borderValue
 @param stream Stream for the asynchronous version.
 
+@note In-place operation is not supported. If src and dst refer to the same data, the behavior is undefined.
 @sa warpAffine
  */
 CV_EXPORTS void warpAffine(InputArray src, OutputArray dst, InputArray M, Size dsize, int flags = INTER_LINEAR,

modules/cudawarping/src/warp.cpp

@@ -208,6 +208,8 @@ void cv::cuda::warpAffine(InputArray _src, OutputArray _dst, InputArray _M, Size
     _dst.create(dsize, src.type());
     GpuMat dst = _dst.getGpuMat();
 
+    CV_Assert( src.data != dst.data && "In-place operation not supported for cv::cuda::warpAffine" );
+
     Size wholeSize;
     Point ofs;
     src.locateROI(wholeSize, ofs);

modules/cudawarping/test/test_warp_affine.cpp

@@ -222,6 +222,36 @@ CUDA_TEST_P(WarpAffine, Accuracy)
     EXPECT_MAT_NEAR(dst_gold, dst, src.depth() == CV_32F ? 1e-1 : 1.0);
 }
 
+CUDA_TEST_P(WarpAffine, OverlapDetection)
+{
+    cv::Mat src = randomMat(size, type);
+    ASSERT_FALSE(src.empty());
+    cv::cuda::GpuMat gpuSrc;
+    gpuSrc.upload(src);
+
+    cv::Mat M = cv::Mat::eye(2, 3, CV_64FC1);
+    int flags = interpolation;
+    if (inverse)
+        flags |= cv::WARP_INVERSE_MAP;
+
+    {
+        cv::cuda::GpuMat gpuDst(gpuSrc, cv::Rect(0, 0, size.width, size.height));
+
+        EXPECT_THROW(
+            cv::cuda::warpAffine(gpuSrc, gpuDst, M, size, flags, borderType, cv::Scalar::all(0)),
+            cv::Exception);
+    }
+
+    {
+        cv::cuda::GpuMat gpuDst(size, gpuSrc.type());
+        ASSERT_NE(gpuSrc.data, gpuDst.data); // Confirm they are distinct
+
+        EXPECT_NO_THROW({
+            cv::cuda::warpAffine(gpuSrc, gpuDst, M, size, flags, borderType, cv::Scalar::all(0));
+        });
+    }
+}
+
 INSTANTIATE_TEST_CASE_P(CUDA_Warping, WarpAffine, testing::Combine(
     ALL_DEVICES,
     DIFFERENT_SIZES,