Normalized covariance needed template arguments

src/c/Common/CudaNormalizedCovariance.cuh

@@ -8,8 +8,8 @@
 template <class PixelType>
 double normalizedCovariance(const PixelType* imageIn1, const PixelType* imageIn2, Vec<size_t> dims, int device=0)
 {
-	double im1Mean = sumArray(imageIn1,dims.product(),device) / (double)dims.product();
-	double im2Mean = sumArray(imageIn2,dims.product(),device) / (double)dims.product();
+	double im1Mean = sumArray<double>(imageIn1,dims.product(),device) / (double)dims.product();
+	double im2Mean = sumArray<double>(imageIn2,dims.product(),device) / (double)dims.product();
 
 	float* im1Sub = addConstant<PixelType,float>(imageIn1,dims,-1.0*im1Mean,NULL,device);
 	float* im2Sub = addConstant<PixelType,float>(imageIn2,dims,-1.0*im2Mean,NULL,device);
@@ -17,14 +17,14 @@ double normalizedCovariance(const PixelType* imageIn1, const PixelType* imageIn2
 	float* im1P = imagePow<float>(im1Sub,dims,2.0,NULL,device);
 	float* im2P = imagePow<float>(im2Sub,dims,2.0,NULL,device);
 
-	double sigma1 = sqrt(sumArray(im1P,dims.product(),device)/(double)dims.product());
-	double sigma2 = sqrt(sumArray(im2P,dims.product(),device)/(double)dims.product());
+	double sigma1 = sqrt(sumArray<double>(im1P,dims.product(),device)/(double)dims.product());
+	double sigma2 = sqrt(sumArray<double>(im2P,dims.product(),device)/(double)dims.product());
 
 	delete[] im1P;
 	delete[] im2P;
 
 	float* imMul = multiplyImageWith<float>(im1Sub,im2Sub,dims,1.0,NULL,device);
-	double numerator = sumArray(imMul,dims.product(),device);
+	double numerator = sumArray<double>(imMul,dims.product(),device);
 
 	double coVar = numerator/(dims.product()*sigma1*sigma2);
 
@@ -37,8 +37,8 @@ double normalizedCovariance(const PixelType* imageIn1, const PixelType* imageIn2
 
 double normalizedCovariance(const double* imageIn1, const double* imageIn2, Vec<size_t> dims, int device=0)
 {
-	double im1Mean = sumArray(imageIn1,dims.product(),device) / (double)dims.product();
-	double im2Mean = sumArray(imageIn2,dims.product(),device) / (double)dims.product();
+	double im1Mean = sumArray<double>(imageIn1,dims.product(),device) / (double)dims.product();
+	double im2Mean = sumArray<double>(imageIn2,dims.product(),device) / (double)dims.product();
 
 	double* im1Sub = addConstant<double,double>(imageIn1,dims,-1.0*im1Mean,NULL,device);
 	double* im2Sub = addConstant<double,double>(imageIn2,dims,-1.0*im2Mean,NULL,device);
@@ -46,14 +46,14 @@ double normalizedCovariance(const double* imageIn1, const double* imageIn2, Vec<
 	double* im1P = imagePow<double>(im1Sub,dims,2.0,NULL,device);
 	double* im2P = imagePow<double>(im2Sub,dims,2.0,NULL,device);
 
-	double sigma1 = sqrt(sumArray(im1P,dims.product(),device)/(double)dims.product());
-	double sigma2 = sqrt(sumArray(im2P,dims.product(),device)/(double)dims.product());
+	double sigma1 = sqrt(sumArray<double>(im1P,dims.product(),device)/(double)dims.product());
+	double sigma2 = sqrt(sumArray<double>(im2P,dims.product(),device)/(double)dims.product());
 
 	delete[] im1P;
 	delete[] im2P;
 
 	double* imMul = multiplyImageWith<double>(im1Sub,im2Sub,dims,1.0,NULL,device);
-	double numerator = sumArray(imMul,dims.product(),device);
+	double numerator = sumArray<double>(imMul,dims.product(),device);
 
 	double coVar = numerator/(dims.product()*sigma1*sigma2);