对于常规的pcb的元件缺失检测大致步骤。
第一：售前，需要快速判断元件是贴片，引脚式，常规裸露式，并使用ROI直接进行快速的blob分析，使用特征面积选择即可快速的完成早期评估工作。
第二：根据需求判断，是否需要模板匹配单对多还是多对多的处理。
第三：实际代码书写，对于模板匹配的输出文件，应以写入新的文件。
本次需求：对c6旁边的电阻进行检测是否存在电阻或者电阻脱落的情况
halcon程序
read_image (Image, 'D:/shijue/11/pcb缺陷/Image__2022-05-04__17-01-48.bmp')
*读图
gen_rectangle2 (ROI_0, 58.5102, 543.681, rad(-101.309), 70.7985, 80.6545)
*对c6进行ROI选取
reduce_domain (Image, ROI_0, ImageReduced2)
*裁图
rgb1_to_gray (ImageReduced2, GrayImage)
*转为单通道灰度值
threshold (GrayImage, Regions2, 168, 220)
*二值化扣出c6
closing_rectangle1 (Regions2, RegionClosing2, 5,5)
*膨胀，防止字符没有扣好
reduce_domain (Image, RegionClosing2, ImageReduced3)
*裁图
inspect_shape_model (ImageReduced3, ModelImages, ModelRegions,5,25)
select_obj (ModelImages, ObjectSelected, 1)
dev_display (ObjectSelected)
create_shape_model (ImageReduced3, 3, 0, rad(360), 'auto', 'none', 'use_polarity', 30, 10, ModelID)
get_shape_model_contours (ModelContours, ModelID, 1)
*以上创建模板
find_shape_model (Image, ModelID, 0, rad(360), 0.7, 1, 0.5, 'least_squares', 0, 0.5, RowCheck, ColumnCheck, AngleCheck, Score)
*根据模板寻找c6
if (|Score| > 0)
*判断是否找到c6
vector_angle_to_rigid (0, 0, 0, RowCheck, ColumnCheck, AngleCheck, HomMat2D)
affine_trans_region (ObjectSelected, RegionAffineTrans, HomMat2D, 'nearest_neighbor')
affine_trans_image (ModelImages, ImageAffineTrans, HomMat2D, 'constant', 'false')
dev_display (ImageAffineTrans)
dev_display (RegionAffineTrans)
*对找到的c6进行仿射变化
orientation_region (RegionAffineTrans, Phi)
area_center (RegionAffineTrans, Area, Row, Column)
*获取c6的坐标和角度
vector_angle_to_rigid (Row, Column, 0, Row, Column, rad(0), HomMat2D)
affine_trans_image (Image, ImageAffineTrans, HomMat2D, 'constant', 'false')
affine_trans_region (RegionAffineTrans, RegionAffineTrans2, HomMat2D, 'nearest_neighbor')
*将c6运用到图中
reduce_domain (ImageAffineTrans, RegionAffineTrans2, ImageReduced4)
*裁图
rgb1_to_gray (ImageReduced4, GrayImage1)
*转为单通道灰度
shape_trans (GrayImage1, RegionTrans, 'convex')
*练成一个区域
area_center (RegionTrans, Area1, Row1, Column1)
orientation_region (RegionTrans, Phi1)
*获取中心坐标和角度
gen_rectangle2_contour_xld (Rectangle1, Row, Column1-300, Phi1, 150,80)
*根据实际绘制一个在c6旁的元件的矩形轮廓
dev_display (ImageAffineTrans)
dev_display (Rectangle1)
*显示检测是否正确
gen_rectangle2 (Rectangle1, Row, Column1-300, Phi1, 150,80)
*根据轮廓，绘制矩形区域
reduce_domain (ImageAffineTrans, Rectangle1, ImageReduced5)
*裁图，对元件进行处理
decompose3 (ImageReduced5, Red, Green, Blue)
trans_from_rgb (Red, Green, Blue, Hue, Saturation, Intensity, 'hsv')
*转化为hsv
threshold (Saturation, Regions, 0, 89)
*对s进行二值化
opening_rectangle1 (Regions, RegionOpening, 3,3)
closing_rectangle1 (RegionOpening, RegionClosing1, 3,3)
*以上为blob分析
connection (RegionClosing1, ConnectedRegions1)
select_shape (ConnectedRegions1, SelectedRegions1, 'area', 'and', 11907.4, 14351.9)
*判断SelectedRegions1数组是否为空即可
endif
*以下为快速实验处理，可在售前阶段，快速评估产品是否有能力完成
*没有元件处理，常规blob分析
gen_rectangle2 (ROI_0, 157, 780.727, rad(-12.3391), 151.872, 74.1818)
reduce_domain (Image, ROI_0, ImageReduced)
decompose3 (ImageReduced, Red, Green, Blue)
trans_from_rgb (Red, Green, Blue, Hue, Saturation, Intensity, 'hsv')
threshold (Saturation, Regions, 0, 89)
opening_rectangle1 (Regions, RegionOpening, 3,3)
closing_rectangle1 (RegionOpening, RegionClosing1, 3,3)
connection (RegionClosing1, ConnectedRegions1)
select_shape (ConnectedRegions1, SelectedRegions1, 'area', 'and', 11907.4, 14351.9)
*有元件处理
read_image (Image1, 'D:/shijue/11/pcb缺陷/Image__2022-04-03__17-25-28.bmp')
gen_rectangle2 (ROI_0, 866.364, 915.182, rad(17.1027), 126.123, 79.6594)
reduce_domain (Image1, ROI_0, ImageReduced1)
decompose3 (ImageReduced1, Red1, Green1, Blue1)
trans_from_rgb (Red1, Green1, Blue1, Hue1, Saturation1, Intensity1, 'hsv')
threshold (Saturation1, Regions1, 0, 81)
opening_rectangle1 (Regions1, RegionOpening1, 3,3)
closing_rectangle1 (RegionOpening1, RegionClosing,3,3)
connection (RegionClosing, ConnectedRegions)
select_shape (ConnectedRegions, SelectedRegions, 'area', 'and', 11907.4, 14351.9)
*判断SelectedRegions数组不为为空即可
实际完成效果
无原件输出面积选择为空
测试的图片参考
由于csdn有要求的图片上传大小，所以这里使用jpg压缩图像并不是原图，有需要原图的可以私信我。
