CvCreateMatND直接訪問

Question

我試過這段代碼，它沒有正確訪問元素，有人可以解釋爲什麼嗎？

int size[] = { 5, 5, 5 }; 
CvMatND* matA = cvCreateMatND(3, size, CV_32F); 
cvSetZero(matA); 


    printf("test: %d \n",cvGetReal3D(matA, 1, 1, 1)); 
    cvSetReal3D(matA, 1, 1, 1, (cvGetReal3D(matA, 1, 1, 1)+1)); 
    int test = cvGetReal3D(matA, 1, 1, 1); 
    printf("test: %d \n",test); 


    char cont = 1; 
     for (int dim_index = 0; dim_index < matA->dims; dim_index++) 
     { 
    printf("There are %d elements on dim %d\n", matA->dim[dim_index].size, dim_index); 

    for (int elem = 0; elem < matA->dim[dim_index].size; elem++) 
    { 
     printf("On element [%d] of dim %d, wrote: %d\n", elem, dim_index, matA->data.ptr[matA->dim[dim_index].size * dim_index + elem]); 
    } 
} 

輸出：

test: 0 
test: 1 
There are 5 elements on dim 0 
On element [0] of dim 0, wrote: 0 
On element [1] of dim 0, wrote: 0 
On element [2] of dim 0, wrote: 0 
On element [3] of dim 0, wrote: 0 
On element [4] of dim 0, wrote: 0 
There are 5 elements on dim 1 
On element [0] of dim 1, wrote: 0 
On element [1] of dim 1, wrote: 0 
On element [2] of dim 1, wrote: 0 
On element [3] of dim 1, wrote: 0 
On element [4] of dim 1, wrote: 0 
There are 5 elements on dim 2 
On element [0] of dim 2, wrote: 0 
On element [1] of dim 2, wrote: 0 
On element [2] of dim 2, wrote: 0 
On element [3] of dim 2, wrote: 0 
On element [4] of dim 2, wrote: 0 

謝謝！

Answer 1

我爲我的其他答案表示歉意，因爲它不正確。你可能在那裏添加了一條評論，說明沒有按預期工作。我有fixed it從那時起，並以正確的方式做你正在嘗試的是：

int size[] = { 5, 5, 5 }; 
CvMatND* matA = cvCreateMatND(3, size, CV_32F); 
cvSetZero(matA); 

printf("Before writting at 1x1x1 %f \n", cvGetReal3D(matA, 1, 1, 1)); 
cvSetReal3D(matA, 1, 1, 1, (cvGetReal3D(matA, 1, 1, 1)+1)); 

int test = cvGetReal3D(matA, 1, 1, 1); 
printf("After: %d\n", test); 

printf("Before writting at 3x2x1 %f \n", cvGetReal3D(matA, 3, 2, 1)); 
cvSetReal3D(matA, 3, 2, 1, (cvGetReal3D(matA, 3, 2, 1)+1)); 

test = cvGetReal3D(matA, 3, 2, 1); 
printf("After: %d\n", test); 

for (int x = 0; x < matA->dim[0].size; x++) // dim 0 
{ 
    for (int y = 0; y < matA->dim[1].size; y++) // dim 1 
    { 
     for (int z = 0; z < matA->dim[2].size; z++) // dim 2 
     { 
      unsigned char* ptr = matA->data.ptr + ((size_t)x * matA->dim[0].step) + ((size_t)y * matA->dim[1].step) + ((size_t)z * matA->dim[2].step); 
      float value = *(float*)ptr; 
      printf("x:%d y:%d z:%d --> %f\n", x, y, z, value); 
     } 
    } 
} 

我提高你的測試一點點，上面的代碼將1寫入位置（1,1,1）和（3 ，2,1）。

輸出（簡體）：

x:1 y:1 z:0 --> 0.000000 
x:1 y:1 z:1 --> 1.000000 
x:1 y:1 z:2 --> 0.000000 
... 
x:3 y:2 z:1 --> 1.000000 
...