cudaarithm: fix the compile faiure of CUDA 12.

The slight API change of NPP nppiMeanStdDevGetBufferHostSize_8u_C1R
The type of bufSize is size_t instead of int in CUDA 12

Author: sdy623

modules/cudaarithm/CMakeLists.txt

@@ -15,6 +15,9 @@ if(ENABLE_CUDA_FIRST_CLASS_LANGUAGE)
     if(NOT CUDA_VERSION VERSION_LESS 10.1)
       list(APPEND optional_dependencies CUDA::cublasLt${CUDA_LIB_EXT})
     endif()
+    if (CUDA_VERSION VERSION_GREATER_EQUAL 12.0)
+      add_definitions(-DCUDA_12_OR_HIGHER)
+    endif()
   endif()
   if(HAVE_CUFFT)
     # static version requires seperable compilation which is incompatible with opencv's current library structure

modules/cudaarithm/src/reductions.cpp

@@ -151,7 +151,12 @@ void cv::cuda::meanStdDev(InputArray src, OutputArray dst, Stream& stream)
     sz.width  = gsrc.cols;
     sz.height = gsrc.rows;
 
+#if (CUDA_VERSION >= 12040)
+    size_t bufSize;  // detail: CUDA>=12 NPP nppi_statistics_functions.h
+#else
     int bufSize;
+#endif
+
 #if (CUDA_VERSION <= 4020)
     nppSafeCall( nppiMeanStdDev8uC1RGetBufferHostSize(sz, &bufSize) );
 #else
@@ -227,7 +232,12 @@ void cv::cuda::meanStdDev(InputArray src, OutputArray dst, InputArray mask, Stre
     sz.width  = gsrc.cols;
     sz.height = gsrc.rows;
 
+#if (CUDA_VERSION >= 12040)
+    size_t bufSize;  // detail: CUDA>=12.4.0 NPP nppi_statistics_functions.h
+#else
     int bufSize;
+#endif
+
 #if (CUDA_VERSION <= 4020)
         nppSafeCall( nppiMeanStdDev8uC1MRGetBufferHostSize(sz, &bufSize) );
 #else