HALCON示例程序color_pieces.hdev通過MLP訓練器對彩色棋子進行分類識別；分別在彩色圖像下與灰度圖像下進行，從而產生對比。

示例程序源碼（加注釋）

• 關于顯示類函數解釋
  dev_update_off ()
  dev_close_window ()
  dev_open_window (0, 0, 557, 416, ‘black’, WindowHandle)
  set_display_font (WindowHandle, 14, ‘mono’, ‘true’, ‘false’)
  dev_set_draw (‘margin’)
• 初始化
  ImageRootName := ‘color/color_pieces_0’
  Regions := [‘yellow’,‘pink’,‘blue’,‘background’]
  Highlight := [‘goldenrod’,‘magenta’,‘cyan’]
  gen_empty_obj (Classes)
  for Mode := 0 to 1 by 1
  dev_set_color (‘black’)
  read_image (Image, ImageRootName + ‘0’)
  • 生成灰度三通道圖像
    if (Mode == 1)
    • rgb1_to_gray - 將RGB圖像轉換為灰度圖像。
    • rgb1_to_gray（RGB圖：灰度圖 ::）；轉換公式：灰色= 0.299 *R+ 0.587 *G+ 0.114 *B。
      rgb1_to_gray (Image, GrayImage)
    • compose3 - 將3個圖像轉換為三通道圖像。
    • compose3（圖1，圖2，圖3：多通道圖像 ::）
      compose3 (GrayImage, GrayImage, GrayImage, Image)
      dev_display (Image)
      disp_message (WindowHandle, ‘Train and apply the classes again on gray images’, ‘window’, 12, 12, ‘black’, ‘false’)
      disp_continue_message (WindowHandle, ‘black’, ‘true’)
      stop ()
      endif
      if (Mode == 0)
    • 指定顏色類
      for I := 1 to 4 by 1
      dev_display (Image)
      dev_display (Classes)
      disp_message (WindowHandle, [‘Drag rectangle inside ’ + Regions[I - 1] + ’ color’,‘Click right mouse button to confirm’], ‘window’, 24, 12, ‘black’, ‘false’)
      draw_rectangle1 (WindowHandle, Row1, Column1, Row2, Column2)
      gen_rectangle1 (Rectangle, Row1, Column1, Row2, Column2)
    • 帶有concat_obj解釋的貼子注意和union1的區別
      concat_obj (Classes, Rectangle, Classes)
      endfor
      endif
  • 創建MLP分類器并添加訓練樣本；關于MLP分類器解釋的例子
    create_class_mlp (3, 7, 4, ‘softmax’, ‘normalization’, 3, 42, MLPHandle)
    add_samples_image_class_mlp (Image, Classes, MLPHandle)
    disp_message (WindowHandle, ‘Training…’, ‘window’, 100, 12, ‘black’, ‘false’)
    train_class_mlp (MLPHandle, 400, 0.5, 0.01, Error, ErrorLog)
    for J := 0 to 3 by 1
    read_image (Image, ImageRootName + J)
    if (Mode == 1)
    rgb1_to_gray (Image, GrayImage)
    compose3 (GrayImage, GrayImage, GrayImage, Image)
    endif
    classify_image_class_mlp (Image, ClassRegions, MLPHandle, 0.5)
    dev_display (Image)
    disp_message (WindowHandle, ‘Looking for 4 game pieces of each color …’, ‘window’, 24, 12, ‘black’, ‘false’)
    dev_set_line_width (2)
    for Figure := 1 to 3 by 1
    * copy_obj - 復制HALCON數據庫中的圖標對象。
    * copy_obj（要復制對象：復制出的對象，開始索引號，對象數量:)
    copy_obj (ClassRegions, ObjectsSelected, Figure, 1)
    * 分割定義域
    connection (ObjectsSelected, ConnectedRegions)
    * 通過面積篩選區域
    select_shape (ConnectedRegions, SelectedRegions, ‘area’, ‘and’, 400, 99999)
    * 對區域進行計數
    count_obj (SelectedRegions, Number)
    dev_set_color (Highlight[Figure - 1])
    dev_display (SelectedRegions)
    OutString := Regions[Figure - 1] + ': ’ + Number + ’ ’
    dev_set_color (‘green’)
    disp_message (WindowHandle, OutString, ‘window’, 24 + 30 * Figure, 12, ‘black’, ‘false’)
    if (Number != 4)
    disp_message (WindowHandle, ‘Not OK’, ‘window’, 24 + 30 * Figure, 120, ‘red’, ‘false’)
    else
    disp_message (WindowHandle, ‘OK’, ‘window’, 24 + 30 * Figure, 120, ‘green’, ‘false’)
    endif
    endfor
    if (J < 3)
    disp_continue_message (WindowHandle, ‘black’, ‘true’)
    stop ()
    endif
    endfor
    endfor
    dev_clear_window ()
    dev_display (Image)
    Message := ‘The game pieces cannot be classified reliable on’
    Message[1] := ‘gray images because the gray values of the’
    Message[2] := ‘game pieces cannot always be distinguished from’
    Message[3] := ‘the gray values of the background.’
    disp_message (WindowHandle, Message, ‘window’, 12, 12, ‘black’, ‘true’)

處理思路

這個例子是將三通道的RGB圖像使用MLP分類器進行分類，分別對灰度圖像與彩色圖像進行了訓練與識別，對比發現還是彩色圖像分類較為準確，因為灰度圖像不能很好地分割出棋子與背景。

后記

大家有什么問題可以向我提問哈，我看到了第一時間回復，希望在學習的路上多多結交良師益友。