halcon——缺陷检测常用方法总结(特征训练)

引言

机器视觉中缺陷检测分为一下几种:

本篇博文主要是对缺陷图像的纹理特征训练进行详细分析。


 特征训练

在纹理中找瑕疵。基于高斯混合模型(GMM)分类器的纹理检查模型,适用于图像金字塔,可以分析纹理的多个频率范围。

要求:训练样本必须完美无瑕疵。

整体步骤:

每层金字塔都会训练一个GMM模型,并确定该层的'novelty_threshold'(区分有无瑕疵的阈值)。

参数获取:get_texture_inspection_model_param

参数设定:set_texture_inspection_model_param

参数分析:'patch_normalization':'weber'对亮度鲁棒,‘none’需要亮度作为评判(默认)

                  'patch_rotational_robustness':'true'对旋转鲁棒,'false'需要旋转作为评判(默认)

                  'levels':设置具体的金字塔层参与训练,纹理越粗糙,则较低的金字塔层级越可省略。默认auto。

                  'sensitivity':灵敏度,影响'novelty_threshold'的计算结果。负值会导致更高的阈值,从而更少的发现缺陷。默认0。

                   'novelty_threshold',阈值,自动计算得到,若结果不理想,可以手动微调。

halcon案例分析(apply_texture_inspection_model.hdev)


 一,创建模型,添加训练样本(完好无损的图像)


TrainingImageIndices := [1,2]
TextureModelFilename := 'texture_model_carpet'
dev_open_window_fit_size (0, 0, Width, Height, -1, -1, WindowHandle1)
dev_display (Image)
*创建模型
 create_texture_inspection_model ('basic', TextureInspectionModel)
 for Index := 0 to |TrainingImageIndices| - 1 by 1
read_image (Image, 'carpet/carpet_' + TrainingImageIndices[Index]$'02')
dev_display (Image)
Message := '添加图片 ' + (Index + 1) + ' of ' + |TrainingImageIndices| + '训练准备'
dev_disp_text (Message, 'window', 12, 12, 'black', [], [])
 *加载训练样本(两张)
 add_texture_inspection_model_image (Image, TextureInspectionModel, Indices)
 endfor


 二,初步设置参数后,开始训练


*参数设定'patch_normalization':'weber'对亮度鲁棒,‘none’需要亮度作为评判(默认)
set_texture_inspection_model_param (TextureInspectionModel, 'patch_normalization', 'weber')
Levels := [2,3,4]
* 'levels':设置具体的金字塔层参与训练,纹理越粗糙,则较低的金字塔层级越可省略。默认auto。
set_texture_inspection_model_param (TextureInspectionModel, 'levels', Levels)
* 开始训练
train_texture_inspection_model (TextureInspectionModel)
*查看样本参数'novelty_threshold',阈值,自动计算得到,若结果不理想,可以手动微调。
get_texture_inspection_model_param (TextureInspectionModel, 'novelty_threshold', NoveltyThreshold)
* 查看各个金字塔等级的新颖性得分图像和新颖性区域,可以把'gen_result_handle'设置为'true',
 *之后get_texture_inspection_result_object读取'novelty_score_image'和'novelty_region'
set_texture_inspection_model_param (TextureInspectionModel, 'gen_result_handle', 'true')


 三,对缺陷图像初测试,显示测试结果


 *设置窗口,用于显示各个金字塔层图像
WindowWidth := 320
WindowHeight := 280
dev_open_window (0, 0, WindowWidth, WindowHeight, 'black', WindowHandle1)
set_display_font (WindowHandle1, 16, 'mono', 'true', 'false')
dev_open_window (0, WindowWidth + 8, WindowWidth, WindowHeight, 'black', WindowHandle2)
set_display_font (WindowHandle2, 16, 'mono', 'true', 'false')
dev_open_window (0, 2 * WindowWidth + 16, WindowWidth, WindowHeight, 'black', WindowHandle3)
set_display_font (WindowHandle3, 16, 'mono', 'true', 'false')
dev_open_window (WindowHeight + 50, WindowWidth / 2 + 8, 2 * WindowWidth, 2 * WindowHeight, 'black', WindowHandle4)
set_display_font (WindowHandle4, 16, 'mono', 'true', 'false')
 WindowHandles := [WindowHandle1,WindowHandle2,WindowHandle3]
  ** 检测第一张训练图像上的纹理缺陷以微调参数。
  for Index := 1 to 3 by 1
    ImageIndex := 5
        read_image (TestImage, 'carpet/carpet_' + ImageIndex$'02')
        *测试当前图像
    apply_texture_inspection_model (TestImage, NoveltyRegion, TextureInspectionModel, TextureInspectionResultID)
* 检查调试信息。
    *查看各个金字塔等级的新颖性得分图像(NovScoreImage)和新颖性区域(NovRegionL)
     * 新颖性评分图像可用于单独微调新颖性阈值。
    get_texture_inspection_result_object (NovScoreImage, TextureInspectionResultID, 'novelty_score_image')
    get_texture_inspection_result_object (NovRegion, TextureInspectionResultID, 'novelty_region')
    * 显示每层(金字塔)的结果
        count_obj (NovScoreImage, Number)
         for Level := 1 to Number by 1
                     CurrentWindow := WindowHandles[Level - 1]
                             dev_set_window (CurrentWindow)
                             dev_clear_window ()
        select_obj (NovScoreImage, NovScoreImageL, Level)
        select_obj (NovRegion, NovRegionL, Level)
        get_image_size (NovScoreImageL, Width, Height)
        dev_set_part (0, 0, Height - 1, Width - 1)
        dev_display (NovScoreImageL)
        Legend := 'Novelty region (level ' + Levels[Level - 1] + ')'
        dev_set_color ('red')
        dev_set_line_width (2)
        * 
        dev_display (NovRegionL)
        dev_disp_text (['Novelty score image (level ' + Levels[Level - 1] + ')','Novelty threshold: ' + NoveltyThreshold[Level - 1]$'.1f'], 'window', 12, 12, 'black', [], [])
        dev_disp_text (Legend, 'window', WindowHeight - 30, 12, 'white', ['box_color','shadow'], ['black','false'])
         endfor
 *显示结果
    dev_set_window (WindowHandle4)
    dev_display (TestImage)
    dev_set_line_width (2)
    dev_set_color ('red')
    dev_display (NoveltyRegion)
    area_center (NoveltyRegion, Area, Row, Column)
    if (Index < 3)
        dev_disp_text ('Result', 'window', 12, 12, 'black', [], [])
    else
        dev_disp_text ('Final result', 'window', 12, 12, 'black', [], [])
    endif


 四,根据测试结果进行微调参数


 * 新奇阈值的微调。
     if (Index == 1)
        Message[0] := '图像中有很多小错误.'
        Message[1] := '可以通过改变 novelty thresholds的值来调整灵敏度(sensitivity—)'
        Message[2] := '例如减少灵敏度参数的值'
        dev_disp_text (Message, 'window', 12, 12, 'black', [], [])

         * 设置阈值计算的灵敏度。 负值导致更高的阈值,因此检测到的缺陷更少。
        * 'sensitivity':灵敏度,影响'novelty_threshold'的计算结果。负值会导致更高的阈值,从而更少的发现缺陷。默认0。
          set_texture_inspection_model_param (TextureInspectionModel, 'sensitivity', -10)
        get_texture_inspection_model_param (TextureInspectionModel, 'novelty_threshold', NoveltyThreshold)
     endif
 if (Index == 2)
        Message := '也可以通过直接操纵新颖性边界来单独调整单个级别的敏感度'
        dev_disp_text (Message, 'window', 12, 12, 'black', [], [])
* 新奇阈值的微调。
         *
         * 从纹理中获取(自动确定的)新奇阈值
         * 检查模型并将适当修改的值设置为新的新颖性阈值。
         *
         *如果我们明确设置新颖性边界,则忽略敏感性。
         * 我们在这里将其重新设置为 0 以避免混淆
        set_texture_inspection_model_param (TextureInspectionModel, 'sensitivity', 0)
        * 
        Offset := [25,10,30]
        get_texture_inspection_model_param (TextureInspectionModel, 'novelty_threshold', NoveltyThreshold)
        set_texture_inspection_model_param (TextureInspectionModel, 'novelty_threshold', Offset + NoveltyThreshold)
        get_texture_inspection_model_param (TextureInspectionModel, 'novelty_threshold', NoveltyThreshold)
    endif
  endfor

for Level := 1 to |WindowHandles| by 1
    dev_set_window (WindowHandles[Level - 1])
    dev_clear_window ()
endfor
dev_set_window (WindowHandle4)
dev_clear_window ()


 五,至此,模型准备完毕,将全部图像进行缺陷检测并显示


*检测所有测试图像上的纹理缺陷。
NumImages := 7
for Index := 1 to NumImages by 1
    read_image (TestImage, 'carpet/carpet_' + Index$'02')
    * 
    *检测当前图像
    apply_texture_inspection_model (TestImage, NoveltyRegion, TextureInspectionModel, TextureInspectionResultID)
    *得到新颖性图像和区域
    get_texture_inspection_result_object (NovScoreImage, TextureInspectionResultID, 'novelty_score_image')
    get_texture_inspection_result_object (NovRegion, TextureInspectionResultID, 'novelty_region')
    * 显示单个金字塔层数的结果
    count_obj (NovScoreImage, Number)
    for Level := 1 to Number by 1
        CurrentWindow := WindowHandles[Level - 1]
        dev_set_window (CurrentWindow)
        dev_clear_window ()
        select_obj (NovScoreImage, NovScoreImageL, Level)
        select_obj (NovRegion, NovRegionL, Level)
        get_image_size (NovScoreImageL, Width, Height)
        dev_set_part (0, 0, Height - 1, Width - 1)
        dev_display (NovScoreImageL)
         Legend := 'Novelty region (level ' + Levels[Level - 1] + ')'
        dev_set_color ('red')
        dev_set_line_width (2)
        * 
        dev_display (NovRegionL)
        dev_disp_text (['Novelty score image (level ' + Levels[Level - 1] + ')','Novelty threshold: ' + NoveltyThreshold[Level - 1]$'.1f'], 'window', 12, 12, 'black', [], [])
        dev_disp_text (Legend, 'window', WindowHeight - 50, 12, ['red','white'], ['box_color','shadow'], ['black','false'])
        endfor
            * 显示结果
             dev_set_window (WindowHandle4)
    dev_display (TestImage)
    dev_set_line_width (2)
    dev_set_color ('red')
    dev_display (NoveltyRegion)
    area_center (NoveltyRegion, Area, Row, Column)
    if (Area > 100)
        dev_disp_text ('Not OK', 'window', 12, 12, 'white', 'box_color', 'red')
    else
        dev_disp_text ('OK', 'window', 12, 12, 'white', 'box_color', 'forest green')
    endif
 if (Index < NumImages)
        dev_disp_text ('Press Run (F5) to continue', 'window', 'bottom', 'right', 'black', [], [])
        stop ()
    endif
endfor

 

【术语解释】

  • Patch:相邻像素的集合。
  • Novelty Score:在测试过程中,将测试图像的纹理特征与纹理检查模型进行比较,并计算它们的'novelty score'。 该值越大,单个纹理特征越不适合纹理检查模型的可能性越大。
  • Novelty Threshold:Novelty Score高于该阈值,则纹理有缺陷。
  • “ novelty_region”是通过组合不同金字塔等级的新颖性区域而生成的,即不同层级金字塔组成的交集区域。如果只有单层金字塔,那么该层的新颖性区域直接就是novelty_region。

    若想查看各个金字塔等级的新颖性得分图像和新颖性区域,可以把'gen_result_handle'设置为'true',之后get_texture_inspection_result_object读取'novelty_score_image'和'novelty_region'。

 

 参考博文:Halcon 纹理缺陷检测 apply_texture_inspection_model - 夕西行 - 博客园 (cnblogs.com)

posted @ 2021-06-16 14:35  唯有自己强大  阅读(5714)  评论(0编辑  收藏  举报