Working bilinear affine warp

Greg Schrock
1 parent e534ff9c
Showing 1 changed file with 85 additions and 9 deletions
openbr/plugins/cuda/cudaaffine.cu
@@ -14,14 +14,88 @@ using namespace cv::gpu;
 namespace br { namespace cuda {
     __device__ __forceinline__ uint8_t cudaaffine_kernel_get_pixel_value(int row, int col, uint8_t* srcPtr, int rows, int cols) {
+        if (row < 0 || row > rows || col < 0 || col > cols) {
+            if (row > rows || col > cols) {
+                return 0;
+            } else{
+                return 0; 
+            }
+        }
+        return (srcPtr + row*cols)[col];
+    }
+
+
+    __device__ __forceinline__ uint8_t cudaaffine_kernel_get_bilinear_pixel_value(double row, double col, uint8_t* srcPtr, int rows, int cols) {
+        // don't do anything if the index is out of bounds
+        if (row < 0 || row > rows || col < 0 || col > cols) {
+            if (row > rows || col > cols) {
+                return 0;
+            } else{
+                return 0; 
+            }
+        }
+        
+        // http://www.sci.utah.edu/~acoste/uou/Image/project3/ArthurCOSTE_Project3.pdf
+        // Bilinear Transformation
+        // f(Px, Py) = f(Q11)×(1−Rx)×(1−Sy)+f(Q21)×(Rx)×(1−Sy)+f(Q12)×(1−Rx)×(Sy)+f(Q22)×(Rx)×(Sy)
+
+        int row1 = floor(row);
+        int row2 = row1+1;
+
+        int col1 = floor(col);
+        int col2 = col1+1;
+
+        double d_row = row - row1;
+        double d_col = col - col1;
+
+        int Q11 = cudaaffine_kernel_get_pixel_value(row1, col1, srcPtr, rows, cols);
+        int Q21 = cudaaffine_kernel_get_pixel_value(row2, col1, srcPtr, rows, cols);
+        int Q12 = cudaaffine_kernel_get_pixel_value(row1, col2, srcPtr, rows, cols);
+        int Q22 = cudaaffine_kernel_get_pixel_value(row2, col2, srcPtr, rows, cols);
+
+        double val = Q22*(d_row*d_col) + Q12*((1-d_row)*d_col) + Q21*(d_row*(1-d_col)) + Q11*((1-d_row)*(1-d_col));
+        return ((uint8_t) round(val));
+    }
+
+    __device__ __forceinline__ uint8_t cudaaffine_kernel_get_distance_pixel_value(double row, double col, uint8_t* srcPtr, int rows, int cols) {
         // don't do anything if the index is out of bounds
         if (row < 1 || row >= rows-1 || col < 1 || col >= cols-1) {
             if (row >= rows || col >= cols) {
                 return 0;
             } else{
-                return 0; }
+                return 0; 
+            }
         }
-        return (srcPtr + row*cols)[col];
+
+        int row1 = floor(row);
+        int row2 = row1+1;
+
+        int col1 = floor(col);
+        int col2 = col1+1;
+
+        double m1 = row2 - row;
+        double m12 = m1*m1;
+
+        double m2 = col - col1;
+        double m22 = m2*m2;
+
+        double d1 = sqrt(m12 - 2*m1 + 1 + m22);
+        double d2 = sqrt(m12 + m22);
+        double d3 = sqrt(m12 - 2*m1 + 1 + m22 - 2*m2 + 1);
+        double d4 = sqrt(m12 + m22 - 2*m2 + 1);
+        double sum = d1 + d2 + d3 + d4;
+
+        double w1 = d1/sum;
+        double w2 = d2/sum;
+        double w3 = d3/sum;
+        double w4 = d4/sum;
+
+        uint8_t v1 = cudaaffine_kernel_get_pixel_value(row1, col1, srcPtr, rows, cols);
+        uint8_t v2 = cudaaffine_kernel_get_pixel_value(row2, col1, srcPtr, rows, cols);
+        uint8_t v3 = cudaaffine_kernel_get_pixel_value(row1, col2, srcPtr, rows, cols);
+        uint8_t v4 = cudaaffine_kernel_get_pixel_value(row2, col2, srcPtr, rows, cols);
+
+        return round(w1*v1 + w2*v2 + w3*v3 + w4*v4);
     }
     /*
@@ -31,9 +105,9 @@ namespace br { namespace cuda {
      * src_row            - The computed source pixel row (mapping from this row)
      * src_col            - The computed source pixel column (mapping from this col)
      */
-    __device__ __forceinline__ void cudaaffine_kernel_get_src_coord(double *trans_inv, int dst_row, int dst_col, int* src_row, int* src_col){
-        *src_col = round(dst_col * trans_inv[0] + dst_row * trans_inv[3] + trans_inv[6]);
-        *src_row = round(dst_col * trans_inv[1] + dst_row * trans_inv[4] + trans_inv[7]);
+    __device__ __forceinline__ void cudaaffine_kernel_get_src_coord(double *trans_inv, int dst_row, int dst_col, double* src_row_pnt, double* src_col_pnt){
+        *src_col_pnt = dst_col * trans_inv[0] + dst_row * trans_inv[3] + trans_inv[6];
+        *src_row_pnt = dst_col * trans_inv[1] + dst_row * trans_inv[4] + trans_inv[7];
         //printf("Dst: [%d, %d, 1] = [%d, %d, 1] \n[ %0.4f, %0.4f, %0.4f] \n[ %0.4f, %0.4f, %0.4f ]\n[ %0.4f, %0.4f, %0.4f ]\n\n", *src_col, *src_row, dst_col, dst_row, trans_inv[0], trans_inv[1], trans_inv[2], trans_inv[3], trans_inv[4], trans_inv[5], trans_inv[6], trans_inv[7], trans_inv[8]);
@@ -47,8 +121,8 @@ namespace br { namespace cuda {
         //printf("Kernel Inv:\n[%0.4f %0.4f %0.4f]\n[%0.4f %0.4f %0.4f]\n[%0.4f %0.4f %0.4f]\n\n", trans_inv[0], trans_inv[1], trans_inv[2], trans_inv[3], trans_inv[4], trans_inv[5], trans_inv[6], trans_inv[7], trans_inv[8]);
-        int srcRowInd;
-        int srcColInd;
+        double srcRowPnt;
+        double srcColPnt;
         // don't do anything if the index is out of bounds
         if (dstRowInd < 1 || dstRowInd >= dst_rows-1 || dstColInd < 1 || dstColInd >= dst_cols-1) {
@@ -60,8 +134,10 @@ namespace br { namespace cuda {
             }
         }
-        cudaaffine_kernel_get_src_coord(trans_inv, dstRowInd, dstColInd, &srcRowInd, &srcColInd);
-        const uint8_t cval = cudaaffine_kernel_get_pixel_value(srcRowInd, srcColInd, srcPtr, src_rows, src_cols); // Get initial pixel value
+        cudaaffine_kernel_get_src_coord(trans_inv, dstRowInd, dstColInd, &srcRowPnt, &srcColPnt);
+        //const uint8_t cval = cudaaffine_kernel_get_distance_pixel_value(srcRowPnt, srcColPnt, srcPtr, src_rows, src_cols); // Get initial pixel value
+        const uint8_t cval = cudaaffine_kernel_get_bilinear_pixel_value(srcRowPnt, srcColPnt, srcPtr, src_rows, src_cols); // Get initial pixel value
+        //const uint8_t cval = cudaaffine_kernel_get_pixel_value(round(srcRowPnt), round(srcColPnt), srcPtr, src_rows, src_cols); // Get initial pixel value
         dstPtr[dstIndex] = cval;
     }