halcon算子報錯1301，但是索引是1的時候能過

del长安 · 发表于 2023-10-17 14:07:39

這是因爲什麽，在索引是1的時候能運行，但是索引變成2以後報錯1301，求大佬指引

del长安 · 发表于 2023-10-17 14:09:48

*********************************************************************************************************
*相機獲取圖片
*open_framegrabber ('USB3Vision', 0, 0, 0, 0, 0, 0, 'progressive', -1, 'default', -1, 'false', 'default', '2BDF59167987_Hikrobot_MVCE12010UM', 0, -1, AcqHandle)
*grab_image_start (AcqHandle, -1)
*grab_image_async (Image, AcqHandle, -1)
*讀取固定圖片
read_image(Image, 'D:/DLtest/testImages/OK/MSIK_0823_165614.bmp')
get_image_size(Image, Width, Height)
dev_open_window(0, 0, Width/4, Height/4, 'black', WindowHandle)
dev_display(Image)
**************************************************
*讀取訓練模型
*理論基礎模型的成功率只有89%，但是實際上239張圖片有10個是錯誤判定，成功率高達96%
*實際成功率比理論高，這個模型是否可以使用
read_dl_model('C:/Users/yifanzhao/Desktop/螺絲訓練/108.hdl',DLModelHandle)
*設置硬件參數
query_available_dl_devices (['runtime', 'runtime', 'id'], ['gpu', 'cpu', 0], DLDeviceHandle)
set_dl_model_param (DLModelHandle, 'device', DLDeviceHandle[0])
*設置預處理參數
create_dl_preprocess_param_from_model (DLModelHandle, 'none', 'full_domain', [], [], [], DLPreprocessParam)
*獲取模型參數
get_dl_model_param (DLModelHandle, 'meta_data', MetaData)
*異常值分類閾值
InferenceClassificationThreshold := number(MetaData.anomaly_classification_threshold)
*閾值下限
InferenceSegmentationThreshold := number(MetaData.anomaly_segmentation_threshold)
*分類字典
DLDatasetInfo := dict{class_names: ['OK', 'NG'], class_ids: [0, 1]}
*窗體字典
WindowDict := dict{}
**************************************************
*排綫
* read_dl_model('C:/Users/yifanzhao/Desktop/螺絲訓練/排綫模型.hdl',DLModelHandlePX)
* query_available_dl_devices(['runtime', 'runtime', 'id'], ['gpu', 'cpu', 0], DLDeviceHandles)
* set_dl_model_param(DLModelHandlePX, 'device', DLModelHandlePX[0])
* create_dl_preprocess_param_from_model(DLModelHandlePX, 'none', 'full_domain', [], [], [], DLPreprocessParamPX)
* get_dl_model_param(DLModelHandlePX, 'meta_data', MetaDataPX)
* InferenceClassificationThreshold:=number(MetaDataPX.anomaly_classification_threshold)
* InferenceSegmentationThreshold := number(MetaDataPX.anomaly_segmentation_threshold)
* WindowDict:=dict{}
* PXOK:=0
* PXNG:=0
**************************************************
*電腦選取
draw_rectangle2(WindowHandle, ComputerRow, ComputerColumn, ComputerPhi, ComputerLength1, ComputerLength2)
gen_rectangle2(ComputerRectangle, ComputerRow, ComputerColumn, ComputerPhi, ComputerLength1, ComputerLength2)
reduce_domain(Image, ComputerRectangle, ImageComputer)
create_shape_model(ImageComputer, 'auto', 0, rad(360), 'auto', 'auto', 'use_polarity', 'auto', 'auto', ModelID)
*標志選取
draw_rectangle2(WindowHandle, SignRow, SignColumn, SignPhi, SignLength1, SignLength2)
gen_rectangle2(SignRectangle, SignRow, SignColumn, SignPhi, SignLength1, SignLength2)
reduce_domain(Image, SignRectangle, ImageSign)
inspect_shape_model(ImageSign, ModelImages, ModelRegions, 4, 30)
create_shape_model(ImageSign, 'auto', 0, rad(360), 'auto', 'auto', 'use_polarity', 'auto', 'auto', ModelID1)
get_shape_model_contours(ModelContours, ModelID1, 1)
area_center(SignRectangle, Area, Row, Column)
*roi選取
gen_empty_obj(RegionRoi)
for Index := 1 to 30 by 1
draw_circle(WindowHandle, circleRow, circleColumn, Radius)
gen_circle(Circle, circleRow, circleColumn, 23)
concat_obj(RegionRoi, Circle, RegionRoi)
endfor
*排綫選取
* gen_empty_obj(PXRoi)
* for IndexPX := 1 to 2 by 1
* draw_rectangle2(WindowHandle, Row3, Column3, Phi, Length1, Length2)
* gen_rectangle2(Rectangle, Row3, Column3, Phi, Length1, Length2)
* concat_obj(PXRoi, Rectangle, PXRoi)
* endfor
*OK與NG標志
resultOK:=0
resultNG:=0
*獲取時間
get_system_time (MSecond, Second, Minute, Hour, Day, YDay, Month, Year)
tuple_string(Month, '.2d', Month_String)
tuple_string(Day, '.2d', Day_String)
tuple_string(Hour, '.2d', Hour_String)
tuple_string(Minute, '.2d', Minute_String)
tuple_string(Second, '.2d', Second_String)
name := 'MSIK_'+Month_String + Day_String + '_'+ Hour_String + Minute_String + Second_String+'.bmp'
*循環驗證
*循環驗證所有螺絲，將每一個位置的螺絲截取為圖片，然後與模型進行對比
*但是DLResult中的數據提取不出來，提取出來就可進行對比。
*是否可以直接將數據拿出或者只是調動數據進行對比。
list_files ('D:/DLtest/testImages/OK', ['files','follow_links'], ImageFiles)
tuple_regexp_select (ImageFiles, ['\\.(tif|tiff|gif|bmp|jpg|jpeg|jp2|png|pcx|pgm|ppm|pbm|xwd|ima|hobj)$','ignore_case'], ImageFiles)
for IndexInference := 0 to |ImageFiles| - 1 by 1
read_image (TestImage, ImageFiles[IndexInference])
*相機獲取圖片
*grab_image(TestImage, AcqHandle)
*讀取固定圖片
*read_image(TestImage, 'D:/DLtest/testImages/OK/MSIK_0823_165614.bmp')
*尋找電腦
find_shape_model(TestImage, ModelID, 0, rad(360), 0.3, 1, 0.5, 'least_squares', 0, 0.9, Row1, Column1, Angle, ComputerScore)
if(|ComputerScore|>0)
*尋找標志物
find_shape_model(TestImage, ModelID1, 0, rad(360), 0.5, 1, 0.5, 'least_squares', 0, 0.9, Row2, Column2, Angle1, SignScore)
if(|SignScore|>0)
*仿射變換匹配標志物位置
vector_angle_to_rigid(0,0,0, Row2, Column2, Angle1, SignHomMat2D)
affine_trans_contour_xld(ModelContours, ContoursAffineTrans, SignHomMat2D)
*仿射變換匹配螺絲ROI位置
vector_angle_to_rigid(Row, Column, 0, Row2, Column2, Angle1, RoiHomMat2D)
affine_trans_region(Circle, RegionAffineTrans, RoiHomMat2D, 'nearest_neighbor')
*仿射變換匹配排綫位置
*vector_angle_to_rigid(0,0,0, Row2, Column2, Angle1, PXHomMat2D)
*affine_trans_region(PXRoi, RegionAffineTransPX, PXHomMat2D, 'nearest_neighbor')
endif
dev_display(TestImage)
*dev_close_window()
*循環檢測
*排綫
*for PXIndex := 1 to 2 by 1
* select_obj(PXRoi, ObjectSelectedPX, PXIndex)
* reduce_domain(TestImage, ObjectSelectedPX, CheckImage)
*endfor
*螺絲
gen_empty_obj(ErrRegion)
for TIndex := 1 to 30 by 1
*搜索螺絲集合的螺絲圖片
select_obj(RegionAffineTrans, ObjectSelected, TIndex)
reduce_domain(TestImage, ObjectSelected, CheckImage)
crop_domain(CheckImage, ImagePart)
*dev_clear_window()
*dev_display(CheckImage)
*圖片與模型檢測
gen_dl_samples_from_images (ImagePart, DLSample)
preprocess_dl_samples (DLSample, DLPreprocessParam)
*分類檢測
apply_dl_model (DLModelHandle, DLSample, [], DLResult)
*判斷閾值
threshold_dl_anomaly_results (InferenceSegmentationThreshold, InferenceClassificationThreshold, DLResult)
*顯示結果
*判斷結果並統計，查看DLResult具體的值并進行分類*********************
dev_display_dl_data (DLSample, DLResult, DLDatasetInfo, ['anomaly_result', 'anomaly_image'], [], WindowDict)
get_dict_param(DLResult, 'keys', [], GenParamValue)
get_dict_tuple(DLResult, 'anomaly_class_id', Tuple)
if(Tuple=0)
resultOK:=resultOK+1
else
resultNG:=resultNG+1
concat_obj(ErrRegion, ImagePart, ErrRegion)
endif
dev_disp_text ('Press F5 (continue)', 'window', 'bottom', 'center', 'black', [], [])
*stop()
endfor
* dev_display(TestImage)
* dev_set_color('green')
* if(resultNG>1)
* dev_set_color('red')
* dev_display(ErrRegion)
* set_display_font (WindowHandle, 30, 'mono', 'true', 'true')
* disp_message (WindowHandle, 'NG', 'window', 12, 12, 'red', 'false')
* wait_seconds(1)
* else
* set_display_font (WindowHandle, 30, 'mono', 'true', 'true')
* disp_message (WindowHandle, 'OK', 'window', 12, 12, 'green', 'false')
* wait_seconds(1)
* endif
stop()
endif
endfor
* while(true)
* endwhile

复制代码

del长安 · 发表于 2023-10-17 14:11:10

del长安发表于 2023-10-17 14:09
*************************************************************************************************** ...

這是代碼，125行的地方在索引為1的時候可以通過，但是索引為2的時候報錯1301了，有沒有大佬解釋一下

Criss · 发表于 2023-10-17 14:13:01

可以统计下RegionAffineTrans这个数量是否存在30个（还是一整个），图形数组的索引从1开始到总数目。

halcon算子報錯1301，但是索引是1的時候能過

浏览过的版块