Zhang-Suen细化算法C#

时间:2013-11-27 14:24:57

标签: c# image algorithm image-processing

我正在尝试按照本指南在C#中编写Zhang-Suen细化算法,而不处理边距。

enter image description here

在功能' zhangsuen'中,我正在阅读图像' imgUndo'并写入图像' img'。 for循环内的指针dataPtrOrigin_aux用于读取3x3窗口内的9个像素,这样dataPtrOrigin_aux5就是该窗口的中心像素,该窗口将沿着整个图像移动,从左到右,从上到下移动。在每个像素中,如果验证语句是否为真,则在指针dataPtrFinal写入的图像中进行相应的更改。

请注意,我将当前像素的邻居存储在8个元素数组中。因此,它们将按照以下顺序存储:

enter image description here

        internal static void zhangsuen(Image<Bgr, byte> img, Image<Bgr, byte> imgUndo)
    {
        unsafe
        {

            MIplImage m = img.MIplImage; //Image to be written.
            MIplImage mUndo = imgUndo.MIplImage; //Image to be read.
            byte* dataPtrFinal = (byte*)m.imageData.ToPointer();
            byte* dataPtrUndo = (byte*)mUndo.imageData.ToPointer();

            int width = img.Width; //Width of the image.
            int height = img.Height; //Height of the image.
            int nChan = m.nChannels; //3 channels (R, G, B).
            int wStep = m.widthStep; //Total width of the image (including padding).
            int padding = wStep - nChan * width; //Padding at the end of each line.

            int x, y, i;

            int[] neighbours = new int[8]; //Store the value of the surrounding neighbours in this array.

            int step; //Step 1 or 2.
            int[] sequence = { 1, 2, 4, 7, 6, 5, 3, 0, 1 };
            int blackn = 0; //Number of black neighbours.
            int numtransitions = 0; //Number of transitions from white to black in the sequence specified by the array sequence.
            int changed = 1; //Just so it enters the while.

            bool isblack = false;

            int counter = 0;


            while(changed > 0)
            {
                changed = 0;

                if (counter % 2 == 0) //We want to read all the pixels in the image before going to the next step
                    step = 1;
                else
                    step = 2;

                for (y = 0; y < height; y++)
                {
                    for (x = 0; x < width; x++)
                    {

                            byte* dataPtrOrigin_aux1 = (byte*)(dataPtrUndo + (y - 1) * m.widthStep + (x - 1) * m.nChannels);
                            byte* dataPtrOrigin_aux2 = (byte*)(dataPtrUndo + (y - 1) * m.widthStep + (x) * m.nChannels);
                            byte* dataPtrOrigin_aux3 = (byte*)(dataPtrUndo + (y - 1) * m.widthStep + (x + 1) * m.nChannels);
                            byte* dataPtrOrigin_aux4 = (byte*)(dataPtrUndo + (y) * m.widthStep + (x - 1) * m.nChannels);
                            byte* dataPtrOrigin_aux5 = (byte*)(dataPtrUndo + (y) * m.widthStep + (x) * m.nChannels);
                            byte* dataPtrOrigin_aux6 = (byte*)(dataPtrUndo + (y) * m.widthStep + (x + 1) * m.nChannels);
                            byte* dataPtrOrigin_aux7 = (byte*)(dataPtrUndo + (y + 1) * m.widthStep + (x - 1) * m.nChannels);
                            byte* dataPtrOrigin_aux8 = (byte*)(dataPtrUndo + (y + 1) * m.widthStep + (x) * m.nChannels);
                            byte* dataPtrOrigin_aux9 = (byte*)(dataPtrUndo + (y + 1) * m.widthStep + (x + 1) * m.nChannels);


                        if (x > 0 && y > 0 && x < width - 1 && y < height - 1)
                        {
                            if (dataPtrOrigin_aux5[0] == 0)
                                isblack = true;

                            if (isblack)
                            {

                                neighbours[0] = dataPtrOrigin_aux1[0];
                                neighbours[1] = dataPtrOrigin_aux2[0];
                                neighbours[2] = dataPtrOrigin_aux3[0];
                                neighbours[3] = dataPtrOrigin_aux4[0];
                                neighbours[4] = dataPtrOrigin_aux6[0];
                                neighbours[5] = dataPtrOrigin_aux7[0];
                                neighbours[6] = dataPtrOrigin_aux8[0];
                                neighbours[7] = dataPtrOrigin_aux9[0];

                                for(i = 0; i <= 7; i++)
                                {
                                    if (neighbours[i] == 0)
                                        blackn++;

                                    if (neighbours[sequence[i]] - neighbours[sequence[i + 1]] == 255) //número de transições de branco para preto, seguindo a ordem do vector sequence
                                        numtransitions++;
                                }


                                if ((blackn >= 2 && blackn <= 6) && numtransitions == 1)
                                {
                                        if (step == 1 && (neighbours[1] == 255 || neighbours[4] == 255 || neighbours[6] == 255) && (neighbours[4] == 255 || neighbours[6] == 255 || neighbours[3] == 255))
                                        {
                                            dataPtrFinal[0] = 255;
                                            dataPtrFinal[1] = 255;
                                            dataPtrFinal[2] = 255;

                                            changed++;
                                        }

                                        if (step == 2 && (neighbours[1] == 255 || neighbours[4] == 255 || neighbours[3] == 255) && (neighbours[1] == 255 || neighbours[6] == 255 || neighbours[3] == 255))
                                        {
                                            dataPtrFinal[0] = 255;
                                            dataPtrFinal[1] = 255;
                                            dataPtrFinal[2] = 255;

                                            changed++;
                                        }

                                }
                            }
                        }


                        dataPtrFinal += nChan;

                        isblack = false;
                        blackn = 0;
                        numtransitions = 0;

                    }
                    dataPtrFinal += padding;

                }


                dataPtrUndo = (byte*)m.imageData.ToPointer(); //Change the image to be read to the one that has just been written.

                counter++;

            }


        }
    }

当我结束阅读第一张图片并将更改写入图片时,我会立即阅读图片。 (一旦(y = 0; y&lt; height; y ++)的循环结束,我希望我刚写的图像是我将在下一个循环中读取的图像,以便进一步细化。我试图完成这与行

dataPtrUndo = (byte*)m.imageData.ToPointer();

虽然计数器的某个值大于0(取决于读取的图像)但是我收到的错误表明已经尝试写入受保护的内存,这表示我已尝试在图像限制之外写入,但我不明白为什么。它是我错误地做的最后一次归因于dataPtrUndo吗?

3 个答案:

答案 0 :(得分:6)

这是我对Zhang-Suen细化算法的C#实现

public static bool[][] ZhangSuenThinning(bool[][] s)
    {
        bool[][] temp = s;
        bool even = true;

        for (int a = 1; a < s.Length-1; a++)
        {
            for (int b = 1; b < s[0].Length-1; b++)
            {
                if (SuenThinningAlg(a, b, temp, even))
                {
                    temp[a][b] = false;
                }
                even = !even;
            }
        }

        return temp;
    }
static bool SuenThinningAlg(int x, int y, bool[][] s, bool even)
    {
        bool p2 = s[x][y - 1];
        bool p3 = s[x + 1][y - 1];
        bool p4 = s[x + 1][y];
        bool p5 = s[x + 1][y + 1];
        bool p6 = s[x][y + 1];
        bool p7 = s[x - 1][y + 1];
        bool p8 = s[x - 1][y];
        bool p9 = s[x - 1][y - 1];


            int bp1 = NumberOfNonZeroNeighbors(x, y, s);
            if (bp1 >= 2 && bp1 <= 6)//2nd condition
            {
                if (NumberOfZeroToOneTransitionFromP9(x, y, s) == 1)
                {
                    if (even)
                    {
                        if (!((p2 && p4) && p8))
                        {
                            if (!((p2 && p6) && p8))
                            {
                                return true;
                            }
                        }
                    }
                    else
                    {
                        if (!((p2 && p4) && p6))
                        {
                            if (!((p4 && p6) && p8))
                            {
                                return true;
                            }
                        }
                    }
                }
            }


        return false;
    }
    static int NumberOfZeroToOneTransitionFromP9(int x, int y, bool[][]s)
    {
        bool p2 = s[x][y - 1];
        bool p3 = s[x + 1][y - 1];
        bool p4 = s[x + 1][y];
        bool p5 = s[x + 1][y + 1];
        bool p6 = s[x][y + 1];
        bool p7 = s[x - 1][y + 1];
        bool p8 = s[x - 1][y];
        bool p9 = s[x - 1][y - 1];

        int A = Convert.ToInt32((p2 == false && p3 == true)) + Convert.ToInt32((p3 == false && p4 == true)) +
                 Convert.ToInt32((p4 == false && p5 == true)) + Convert.ToInt32((p5 == false && p6 == true)) +
                 Convert.ToInt32((p6 == false && p7 == true)) + Convert.ToInt32((p7 == false && p8 == true)) +
                 Convert.ToInt32((p8 == false && p9 == true)) + Convert.ToInt32((p9 == false && p2 == true));
        return A;
    }
    static int NumberOfNonZeroNeighbors(int x, int y, bool[][]s)
    {
        int count = 0;
        if (s[x-1][y])
            count++;
        if (s[x-1][y+1])
            count++;
        if (s[x-1][y-1])
            count++;
        if (s[x][y+1])
            count++;
        if (s[x][y-1])
            count++;
        if (s[x+1][y])
            count++;
        if (s[x+1][y+1])
            count++;
        if (s[x+1][y-1])
            count++;
        return count;
    }

答案 1 :(得分:6)

bwang22的答案有效。有点。但有两个问题:它不会进行迭代,直到不再发生更改。它做了一个浅层的阵列..两个问题合作可以说,相互抵消,导致变薄,但不是最好看的。

这是更正后的代码,它可以提供更好看的结果:

从Image转换为bool [] []并返回的前两种方法;功能没有针对速度进行优化;如果你需要那个去lockbits / unsafe ..:

public static bool[][] Image2Bool(Image img)
{
    Bitmap bmp = new Bitmap(img);
    bool[][] s = new bool[bmp.Height][];
    for (int y = 0; y < bmp.Height; y++ )
    {
        s[y] = new bool[bmp.Width];
        for (int x = 0; x < bmp.Width; x++)
            s[y][x] = bmp.GetPixel(x, y).GetBrightness() < 0.3;
    }
    return s;

}

public static Image Bool2Image(bool[][] s)
{
    Bitmap bmp = new Bitmap(s[0].Length, s.Length);
    using (Graphics g = Graphics.FromImage(bmp)) g.Clear(Color.White);
    for (int y = 0; y < bmp.Height; y++)
        for (int x = 0; x < bmp.Width; x++)
            if (s[y][x]) bmp.SetPixel(x, y, Color.Black);

    return (Bitmap)bmp;
}

现在已修正的细化代码,其中大部分或多或少与bwang22的答案不变:

public static bool[][] ZhangSuenThinning(bool[][] s)
{
    bool[][] temp = ArrayClone(s);  // make a deep copy to start.. 
    int count = 0;
    do  // the missing iteration
    {
        count  = step(1, temp, s);
        temp = ArrayClone(s);      // ..and on each..
        count += step(2, temp, s);
        temp = ArrayClone(s);      // ..call!
    }
    while (count > 0);

    return s;
}

static int step(int stepNo, bool[][] temp, bool[][] s)
{
    int count = 0;

    for (int a = 1; a < temp.Length - 1; a++)
    {
        for (int b = 1; b < temp[0].Length - 1; b++)
        {
            if (SuenThinningAlg(a, b, temp, stepNo == 2))
            {
                // still changes happening?
                if (s[a][b]) count++; 
                s[a][b] = false; 
            }
        }
    }
    return count;
}

static bool SuenThinningAlg(int x, int y, bool[][] s, bool even)
{
    bool p2 = s[x][y - 1];
    bool p3 = s[x + 1][y - 1];
    bool p4 = s[x + 1][y];
    bool p5 = s[x + 1][y + 1];
    bool p6 = s[x][y + 1];
    bool p7 = s[x - 1][y + 1];
    bool p8 = s[x - 1][y];
    bool p9 = s[x - 1][y - 1];


    int bp1 = NumberOfNonZeroNeighbors(x, y, s);
    if (bp1 >= 2 && bp1 <= 6) //2nd condition
    {
        if (NumberOfZeroToOneTransitionFromP9(x, y, s) == 1)
        {
            if (even)
            {
                if (!((p2 && p4) && p8))
                {
                    if (!((p2 && p6) && p8))
                    {
                        return true;
                    }
                }
            }
            else
            {
                if (!((p2 && p4) && p6))
                {
                    if (!((p4 && p6) && p8))
                    {
                        return true;
                    }
                }
            }
        }
    }
    return false;
}

static int NumberOfZeroToOneTransitionFromP9(int x, int y, bool[][] s)
{
    bool p2 = s[x][y - 1];
    bool p3 = s[x + 1][y - 1];
    bool p4 = s[x + 1][y];
    bool p5 = s[x + 1][y + 1];
    bool p6 = s[x][y + 1];
    bool p7 = s[x - 1][y + 1];
    bool p8 = s[x - 1][y];
    bool p9 = s[x - 1][y - 1];

    int A = Convert.ToInt32((!p2  && p3 )) + Convert.ToInt32((!p3  && p4 )) +
            Convert.ToInt32((!p4  && p5 )) + Convert.ToInt32((!p5  && p6 )) +
            Convert.ToInt32((!p6  && p7 )) + Convert.ToInt32((!p7  && p8 )) +
            Convert.ToInt32((!p8  && p9 )) + Convert.ToInt32((!p9  && p2 ));
    return A;
}
static int NumberOfNonZeroNeighbors(int x, int y, bool[][] s)
{
    int count = 0;
    if (s[x - 1][y])     count++;
    if (s[x - 1][y + 1]) count++;
    if (s[x - 1][y - 1]) count++;
    if (s[x][y + 1])     count++;
    if (s[x][y - 1])     count++;
    if (s[x + 1][y])     count++;
    if (s[x + 1][y + 1]) count++;
    if (s[x + 1][y - 1]) count++;
    return count;
}

我保留了原始的偶数标记,但是通过比较步骤编号来调用它。我直接使用bools保存了几个字符..

最后一个函数来获取嵌套的2d数组的深层副本:

public static T[][] ArrayClone<T>(T [][] A) 
     { return A.Select(a => a.ToArray()).ToArray(); }

这是使用两个PictureBoxes来调用它的方法:

pictureBox1.Image = Image.FromFile("D:\\RCdemo.png");
bool[][] t = Image2Bool(pictureBox1.Image);
t = ZhangSuenThinning(t);
pictureBox2.Image = Bool2Image(t);

我附上一张测试图片。

thinning screenshot

答案 2 :(得分:0)

bwang22的答案很慢。尝试以下方法:

public readonly struct ConnectivityData
{
    public readonly int[] N;
    public readonly int NumNeighbors;
    public readonly int NumChanges;

    public ConnectivityData(in int[] n, in int numNeighbors, in int numChanges)
    {
        N = n;
        NumNeighbors = numNeighbors;
        NumChanges = numChanges;
    }
}


public static void ZhangSuen(in HashSet<Pixel> pixels)
{        
    while (true)
    {
        // Pass #1:
        List<Pixel> mark1 = new List<Pixel>();
        foreach (Pixel p in pixels)
        {
            ConnectivityData conn = ComputeConnectivity(p, pixels);

            if (conn.NumNeighbors > 1 && 
                conn.NumNeighbors < 7 && 
                conn.NumChanges == 1 &&
                conn.N[0] * conn.N[2] * conn.N[4] == 0 &&
                conn.N[2] * conn.N[4] * conn.N[6] == 0)
            {
                mark1.Add(p);
            }
        }

        //delete all marked:
        foreach (Pixel p in mark1)
        {
            pixels.Remove(p);
        }

        // PASS #2:
        List<Pixel> mark2 = new List<Pixel>();
        foreach (Pixel p in pixels)
        {
            ConnectivityData conn = ComputeConnectivity(p, pixels);

            if (conn.NumNeighbors > 1 &&
                conn.NumNeighbors < 7 &&
                conn.NumChanges == 1 &&
                conn.N[0] * conn.N[2] * conn.N[6] == 0 &&
                conn.N[0] * conn.N[4] * conn.N[6] == 0)
            {
                mark2.Add(p);
            }
        }

        //delete all marked:
        foreach (Pixel p in mark2)
        {
            pixels.Remove(p);
        }

        if (mark1.Count == 0 && mark2.Count == 0)
        {
            break;
        }
    }

}

private static ConnectivityData ComputeConnectivity(
    in Pixel p, 
    in HashSet<Pixel> pixels)
{
    // calculate #neighbors and number of changes:
    int[] n = new int[8];
    if (pixels.Contains(new Pixel(p.X, p.Y - 1)))
    {
        n[0] = 1;
    }
    if (pixels.Contains(new Pixel(p.X + 1, p.Y - 1)))
    {
        n[1] = 1;
    }
    if (pixels.Contains(new Pixel(p.X + 1, p.Y)))
    {
        n[2] = 1;
    }
    if (pixels.Contains(new Pixel(p.X + 1, p.Y + 1)))
    {
        n[3] = 1;
    }
    if (pixels.Contains(new Pixel(p.X, p.Y + 1)))
    {
        n[4] = 1;
    }
    if (pixels.Contains(new Pixel(p.X - 1, p.Y + 1)))
    {
        n[5] = 1;
    }
    if (pixels.Contains(new Pixel(p.X - 1, p.Y)))
    {
        n[6] = 1;
    }
    if (pixels.Contains(new Pixel(p.X - 1, p.Y - 1)))
    {
        n[7] = 1;
    }

    return new ConnectivityData(
            n,
            n[0] + n[1] + n[2] + n[3] + n[4] + n[5] + n[6] + n[7],
            ComputeNumberOfChanges(n));
}

private static int ComputeNumberOfChanges(in int[] n)
{
    int numberOfChanges = 0;

    // Iterate over each location and see if it is has changed from 0 to 1:
    int current = n[0];
    for (int i = 1; i < 8; i++)
    {
        if (n[i] == 1 && current == 0)
        {
             numberOfChanges++;
        }
        current = n[i];
    }

    // Also consider the change over the discontinuity between n[7] and n[0]:
    if (n[0] == 1 && n[7] == 0)
    {
        numberOfChanges++;
    }

    return numberOfChanges;
}

要使用:

在您的位图等中,创建一个类型为Pixel的哈希集(包含您要稀疏的所有黑色像素),例如:

public class Pixel
{
    public int X;
    public int Y;

    public Pixel(in int x, in int y)
    {
        X = x;
        Y = y;
    }

    public override bool Equals(object pixel)
    {
        Pixel b = pixel as Pixel;
        return X == b.X && Y == b.Y;
    }

    public override int GetHashCode()
    {
        //return (a.X << 2) ^ a.Y; // this is also commonly used as a pixel hash code
        return X * 100000 + Y; // a bit hacky [will fail if bitmap width is > 100000]
    }
}

...然后调用ZhangSuen(pixels)。这将从集合中删除适当的像素。

请注意,此方法不适用于所有图像。它会使部分图像消失。具体来说,我遇到的问题是右下角的对角线的粗细约为11像素宽。

我目前正在研究一种改进方法,但是在我测试过的大多数文件(CAD文件)中,它的性能都优于类似的斯坦尼福德算法。