图像基础---二值图像的轮廓提取

二值图像的轮廓提取对于图像识别,图像分割有着重要意义。该算法的核心就是将图像目标的内部点消除。所谓内部点,我们要根据当前像素点的邻域来进行判断,假设邻域窗口为3*3窗口,如果当前像素P(x,y)的八个邻域像素满足如下条件,则该点即内部点:

  1P(x,y)为目标像素,假设目标像素为黑色0,背景像素为白色255,那么P(x,y)=0;

  2P(x,y)的八个邻域像素均为目标像素0

  我们把满足条件的内部点删除,换为背景点255,即可得到图像轮廓。

  内部点如下图所示:

 Fig.1内部点

[函数代码]

        /// 

        /// Contour Extraction process.

        /// 

        /// The source image.

        /// 

        public static WriteableBitmap ContourExtraction(WriteableBitmap src)

        {

            if (src != null)

            {

                int w = src.PixelWidth;

                int h = src.PixelHeight;

                WriteableBitmap srcImage = new WriteableBitmap(w, h);

                byte[] temp = src.PixelBuffer.ToArray();

                byte[] tempMask = (byte[])temp.Clone();

                for (int j = 1; j < h-1; j++)

                {

                    for (int i = 4; i < w*4-4; i+=4)

                    {

                        if ((tempMask[i + j * w * 4] == 0) && (tempMask[i - 4 + j * w * 4] == 0) && (tempMask[i + 4 + j * w * 4] == 0) && (tempMask[i - 4 + (j - 1) * w * 4] == 0)

                            && (tempMask[i - 4 + (j + 1) * w * 4] == 0) && (tempMask[i + (j - 1) * w * 4] == 0) && (tempMask[i + (j + 1) * w * 4] == 0)

                            && (tempMask[i + 4 + (j - 1) * w * 4] == 0) && (tempMask[i + 4 + (j + 1) * w * 4] == 0))

                        {

                            temp[i + j * w * 4] = (byte)255;

                            temp[i + 1 + j * w * 4] = (byte)255;

                            temp[i + 2 + j * w * 4] = (byte)255;

                        }

                    }

                }

                Stream sTemp = srcImage.PixelBuffer.AsStream();

                sTemp.Seek(0, SeekOrigin.Begin);

                sTemp.Write(temp, 0, w * 4 * h);

                return srcImage;

            }

            else

            {

                return null;

            }

        }

[图像效果]

Fig.2二值图像                      Fig.3轮廓图像

展开阅读全文

没有更多推荐了,返回首页