基于Aforge的手势识别之一 简单的手写识别

               

本文来自http://blog.youkuaiyun.com/hellogv/ ，引用必须注明出处！

     上一篇文章介绍了如何用Aforge去捕捉运动物体，现在就介绍一个更深入的操作----手势识别。
      我实现手势识别的原理很简单：捕捉运动物体+手写识别，把运动的物体的轨迹记录下来，然后通过手写识别引擎去搜索数据中最匹配的数据，从而知道“写”的是什么。目前常见的开源手写识别引擎有zinnia，wagomu 这些，不过小弟我比较业余，只把网上的比较常见的手写识别代码改进一下，只能识别字母和数字，真想通过摄像头隔空“手写”的朋友就要多花时间玩玩上面提到的几个开源手写类库了。

      本文介绍的手写识别：先在一个固定大小的画板上，用鼠标画下某图形，输入该图形对应的字母，程序把画板上的字母特征点都保存下来特征数据库(相当于学习记忆)，然后再在画板上画出类似该字母的图形，程序就通过新画的特征点搜索特征数据库从而找出最类似的字母。

接下来贴出核心代码，详细的代码请到这里下载：http://download.youkuaiyun.com/source/2312865

GetBMPContext()是把画板中的图形的特征分析出来，Learn()是把特征与特定的字母/数字对应起来保存到数据库，Recognise()是把当前画板的图形特征从数据库中搜索，从而找出对应的字母/数字。

        const int SCAN_GAP = 10;        private String GetBMPContext(Bitmap bmp)        {            Boolean bool1stScan = true;            int ax = 0, ay = 0, bx = 0, by = 0;            String result = "";            for (int i = 1; i < bmp.Width; i = i + SCAN_GAP)            {                for (int j = 1; j < bmp.Height; j = j + SCAN_GAP)                {                    if (bmp.GetPixel(i, j).ToArgb() == Color.Black.ToArgb())                    {                        if (bool1stScan == false)                        {                            if (i <= ax) ax = i;                            if (i >= bx) bx = i;                            if (j <= ay) ay = j;                            if (j >= by) by = j;                        }                        else                        {                            bool1stScan = false;                            ax = i;                            bx = i;                            ay = j;                            by = j;                        }                    }                }            }            Bitmap bmp2 = new Bitmap(20, 20);            Graphics g2 = Graphics.FromImage((Image)bmp2);            g2.Clear(Color.White);            g2.DrawImage(bmp, new Rectangle(0, 0, bmp2.Width, bmp2.Height),                new Rectangle(0, 0, bmp.Width, bmp.Height), GraphicsUnit.Pixel);            g2.Dispose();            //            pictureBox1.Image = bmp2;            int a = 0, b = 0;            for (int i = 0; i < bmp2.Width; i++)            {                for (int j = 0; j < bmp2.Height; j++)                {                    if (bmp2.GetPixel(i, j).ToArgb() == Color.Black.ToArgb())                        result = result + "0";                    else                        result = result + "1";                }            }            return result;        }        public void Learn(String name)        {            StreamWriter sw = new StreamWriter(fileName, true);            sw.WriteLine(name + " " + GetBMPContext(bmpDraw));            sw.Close();        }        public String Recognise()        {            String current = GetBMPContext(bmpDraw);            StreamReader sr = new StreamReader(fileName);            int max = 0;            String result = "";            while (sr.EndOfStream == false)            {                String[] key = sr.ReadLine().Split(' ');                String name = key[0];                String data = key[1];                int match = 0;                for (int i = 0; i < current.Length; i++)                {                    if (current[i] == data[i])                        match++;                }                if (match >= max)                {                    result = name;                    max = match;                }                //Trace.WriteLine(result + ":" + match + "," + max);            }            sr.Close();            return result;        }  

           

再分享一下我老师大神的人工智能教程吧。零基础！通俗易懂！风趣幽默！还带黄段子！希望你也加入到我们人工智能的队伍中来！https://blog.youkuaiyun.com/jiangjunshow