Question

我使用OpenCVSharp（OpenCvSharp3-AnyCPU版本3.0.0.20150823在Visual Studio 2015中运行并通过NuGet安装）从C＃访问OpenCV，但在调用Cv2.PCACompute时，我得到了一个通用 OpenCVException 声明我有一个不支持的输入和输出数组格式组合。

我的目标是使用PCA查找像素blob的主轴。这是我目前的（剥离）代码：

using OpenCvSharp;

public struct point2D
{
     public int X;
     public int Y;

     public point2D(int X, int Y)
     {
          this.X = X;
          this.Y = Y;
     }
}

public static void PCA2D()
{
    int height = 5;
    int width = 5;

    int[] image = new int[]
    {
         0,0,0,0,1,
         0,0,0,1,0,
         0,0,1,0,0,
         0,1,0,0,0,
         1,0,0,0,0,
    }

    // extract the datapoints
    List<point2D> dataPoints = new List<point2D>();

    for (int row = 0; row < height; ++row)
    {
         for (int col = 0; col < width; ++col)
         {
              if (image[row * width + col] == 1)
              {
                  dataPoints.Add(new point2D(col, row));
              }
         }
    }

    // create the input matrix
    Mat input = new Mat(dataPoints.Length, 2, MatType.CV_32SC1);

    for (int i = 0; i < dataPoints.Length; ++i)
    {
        input.Set(i, 0, dataPoints[i].X);
        input.Set(i, 1, dataPoints[i].Y);
    }

    Mat mean = new Mat();
    Mat eigenvectors = new Mat();

    // OpenCVException occurs here: unsupported combination of input and output array formats
    Cv2.PCACompute(input, mean, eigenvectors);

    // Code to get orientation from the eigenvectors
}

我还没有找到关于如何初始化均值和特征向量Mats的文档，或者我调用PCACompute的方式是否正确。对使用PCACompute的正确程序有一些了解会非常有帮助。

Answer 1

事实证明dataPoints不能是MatType.CV_32SC1。将代码更改为以下代码可以使其工作：

// create the input matrix
Mat input = new Mat(dataPoints.Length, 2, MatType.CV_32FC1);

for (int i = 0; i < dataPoints.Length; ++i)
{
    input.Set(i, 0, (float)dataPoints[i].X);
    input.Set(i, 1, (float)dataPoints[i].Y);
}

OpenCvSharp PCA异常：不支持的输入和输出数组格式组合

1 个答案: