Halcon数据类型转换系列(1)图像image、区域region和轮廓xld的相互转换(★firecat推荐★)

本篇记录halcon图像image、区域region和轮廓xld的相互转换

一、xld转region
*不裁剪画布外面的部分
set_system ('clip_region', 'false')

方法1:

gen_region_contour_xld (SelectedXLD, RegionXLD, 'filled')

方法2:

*UnionContours1为轮廓
sort_contours_xld (UnionContours1, SortedContours, 'upper_left', 'true', 'column')
count_obj (UnionContours1, Number)
*将轮廓转换成区域
gen_empty_obj (Line)
for i := 1 to Number by 1
select_obj (SortedContours, ObjectSelected, i)
get_contour_xld (ObjectSelected, Row, Col)
gen_region_polygon (Region, Row, Col)
concat_obj (Line, Region, Line)
endfor
方法3:

gen_region_polygon_filled (Region1, Row, Col)

gen_contour_polygon_xld (Contour, Row, Col)
*判断轮廓是不是闭合
test_closed_xld (Contour, IsClosed)
if (IsClosed == 1)
endif
close_contours_xld (Contour, ClosedContours)
gen_region_contour_xld (ClosedContours, Region2, 'filled')

二、region转xld
方法1:gen_contour_region_xld (SelectedRegions, Contours, 'border')
拟合部分边缘提取和轮廓分割之间会用到,因为轮廓分割需输入xld轮廓,而用boundary提取区域边缘输出的是区域(region),所以需要转换。

方法2:先将区域转换骨架,然后再提取骨架轮廓

skeleton (Region2, Skeleton2)
gen_contours_skeleton_xld (Skeleton2, Contours, 1, 'filter')

三、xld/region转换成image
方法1:

*Halcon感兴趣区域填充特定颜色
color24 := [255,0,0]
color8 := 255
gen_region_contour_xld (UnionContoursCircles, Region1, 'filled')

*区域转换为图片
region_to_bin(Region, Binary, 0, 255, Width, Height)
overpaint_region (Binary, Region1, color8, 'fill')
write_image (Binary, 'bmp', 0, 'E:/Org.bmp')

或者

threshold (Image1, Region, 128, 255)
region_to_bin(Region, Binary, 0, 255, Width, Height)
write_image (Binary, 'bmp', 0, 'E:/Org.bmp')
或者
binary_threshold (Image1, BrightRegion, 'max_separability', 'dark', UsedThreshold)
region_to_bin(BrightRegion, SaveBinary, 0, 255, Width, Height)
write_image (SaveBinary, 'jpeg 100', 0, 'd:/Org.jpg')

方法2:

*创建空白图像,将得到的区域贴到上面
get_image_size (ImageReduced, Width1, Height1)
gen_image_proto (ImageReduced, ImageCleared, 128)
paint_region (Region, ImageCleared, ImageResult1, 255, 'fill')

方法3:(推荐)xld->region->image

*无效set_system ('init_new_image', 'false')
gen_region_contour_xld (ObjectSelected, Region, 'filled')
gen_image_const (NewImage, 'byte', Width, Height)
*Create an image with a specified constant gray value
gen_image_proto (NewImage, ImageCleared1, 255)
*Paint regions into an image
paint_region (Region, ImageCleared1, ImageResult, 0, 'fill')

write_image (ImageResult, 'jpeg', 0, 'D:/1111.jpg')

*Overpaint regions in an image
gen_image_proto (NewImage, ImageCleared2, 255)
overpaint_region(ImageCleared2, Region, 0, 'fill')

结论:paint_region 和overpaint_region最终的输出结果是一样的

四、从image裁剪需要的区域,成为新的image
read_image(Image,'monkey')

gen_rectangle1 (ROI_0, 588.03, 468.95, 2328.43, 3212.37)
reduce_domain (Image, ROI_0, ImageReduced)

crop_domain(ImageReduced, ImagePart)

write_image(ImagePart, 'bmp', 0, 'e:/1.bmp')

五、从image获得region
get_domain (Image, Domain)

binary_threshold (Image, Region, 'smooth_histo', 'dark', UsedThreshold)

mean_image (Image, ImageMean, 12, 12)
dyn_threshold (Image, ImageMean, Region, 30, 'dark')

read_image (Image2, 'D:/1.jpg')
draw_rectangle1 (WindowHandle, Row1, Column1, Row2, Column2)
gen_rectangle1 (Rectangle, Row1, Column1, Row2, Column2)
reduce_domain (Image2, Rectangle, ImageReduced)
threshold (ImageReduced, Region, 100, 255)
area_center (Region, Area3, Row3, Column3)
for I := 0 to |Row3| - 1 by 1
gen_cross_contour_xld (Cross, Row3[I], Column3[I], 20, 0)
endfor
stop()

六、综合

*动态二值化
D:=31
mean_image(Image1, Mean, D*2+1, D*2+1)
dyn_threshold(Image1, Mean, Seg, 5,'dark')
R:=2.5
erosion_circle (Seg, RegionErosion, R)
dilation_circle (RegionErosion, RegionDilation, R)
fill_up (RegionDilation, RegionFillUp)
connection(RegionFillUp, Regions)
select_shape (Regions, SelectedRegions, 'area', 'and', 100, 30000)
region_to_bin(SelectedRegions, Binary, 0, 255, Width, Height)
area_center (SelectedRegions, Area, Row, Column)

for I := 0 to |Row| - 1 by 1
gen_cross_contour_xld (Cross, Row[I], Column[I], 20, 0)
endfor
stop()


————————————————
版权声明:本文为CSDN博主「libaineu2004」的原创文章,遵循CC 4.0 BY-SA版权协议,转载请附上原文出处链接及本声明。
原文链接:https://blog.csdn.net/libaineu2004/article/details/102828588

 

posted @ 2022-03-05 09:48  专注视觉  阅读(747)  评论(0编辑  收藏  举报