halcon入门之_提取遥控器字符并且写入txt文本

程序中使用的照片如上

*老生常谈
read_image (Image, a)
rgb1_to_gray (Image, GrayImage)
dev_display (GrayImage)
* 首先要把它转化为灰度图,在机器视觉中,大部分的图像处理算子都是建立在灰度图上的,所以gray(灰度图)是标志性存在的*转化为灰度图后,就要进入正式的图像处理了,先来一波阈值分割auto_threshold (GrayImage, Regions, 24.4)
*auto_threshold 算子是基于直方图的自动阈值分割方法,第三个参数的作用是一个平滑算子,在这次的使用中,效果不是很好,分析原因主要是对比不够强烈,算子使用前后没有明显区分
threshold (GrayImage, Region, 170, 255)
*threshold 是手动设置阈值的算子,在halcon中查看阈值可以在鼠标移至照片上,按下Ctrl键,便会实时显示当前灰度值
*binary_threshold (GrayImage, Region, 'smooth_histo', 'light', UsedThreshold)
*binary_threshold 在这里的使用中,遥控器左下角有些许噪声会干扰
*阈值分割后,常见的做法就是连通区域
connection (Region, ConnectedRegions)
*然后在这里,明显的可以通过面积特征来筛选就可以
select_shape_std (ConnectedRegions, SelectedRegions, 'max_area', 1.22555e+06)
dev_clear_window ()
dev_display (SelectedRegions)*在select_shape_std 算子中,面积大小可以通过halcon的"打开特征检测"菜单来查看*提取之后填充
fill_up (SelectedRegions, RegionFillUp)
*截取带有字符的区域,如果上边不填充,这里直接截取区域的话,会漏掉字符部分
reduce_domain (GrayImage, RegionFillUp, ImageReduced)
*区域截取之后,再通过阈值分割来找到字符
fast_threshold (ImageReduced, Region1, 0, 100, 20)
*先把字符区域放平整,分割出去','
orientation_region (Region1, Phi)
*转换成正常角度,把逗号分割出去
area_center (Region1, Area, Row, Column)
vector_angle_to_rigid (Row, Column, Phi, Row, Column, -0.5*3.14, HomMat2D)
affine_trans_region (Region1, RegionAffineTrans, HomMat2D, 'nearest_neighbor')*去除','区域,在这里考虑使用面积筛选
*腐蚀-膨胀-连通处理*筛选
closing_circle (RegionAffineTrans, RegionClosing, 3.5)
dilation_rectangle1 (RegionClosing, RegionDilation, 1, 11)
connection (RegionDilation, ConnectedRegions1)
******这里的两个筛选还是有区别的,注意区分
select_shape_std (ConnectedRegions1, SelectedRegions1, 'max_area', 150)
select_shape (ConnectedRegions1, SelectedRegions2, 'area', 'and', 150, 99999)
****把未去除逗号区域旋转
affine_trans_region (Region1, RegionAffineTrans1, HomMat2D, 'nearest_neighbor')
*求交集
intersection (SelectedRegions2, RegionAffineTrans1, RegionIntersection)sort_region (RegionIntersection, SortedRegions, 'first_point', 'true', 'row')
*有字符连接在一起的情况,要分割开来
*首先寻找特征,根据面积呢,是不合适了,在这里,我们选择通过矩形边长来吧select_shape (SortedRegions, SelectedRegions3, 'rect2_len1', 'and', 30, 99999)
*膨胀
closing_circle (SelectedRegions3, RegionClosing1, 0.5)
dilation_rectangle1 (RegionClosing1, RegionDilation1, 1, 11)
connection (RegionDilation1, ConnectedRegions2)*单个字符分割后,再求交集
intersection (ConnectedRegions2, SelectedRegions3, RegionIntersection1)*得到一个空区域
gen_empty_obj (EmptyObject)
concat_obj (EmptyObject, RegionIntersection1, EmptyObject)difference (RegionIntersection, RegionIntersection1, RegionDifference)concat_obj (RegionDifference, EmptyObject, ObjectsConcat)count_obj (ObjectsConcat, Number)a:=[]
l1:=[]
l2:=[]
for i := 1 to Number by 1select_obj (ObjectsConcat, ObjectSelected, i)area_center (ObjectSelected, Area1, Row1, Column1)a:=[a,Area1,'\t']smallest_rectangle2 (ObjectSelected, Row2, Column2, Phi1, Length1, Length2)l1:=[l1,Length1,'\t']l2:=[l2,Length2,'\t']
endfor*新建txt
open_file ('0711.txt', 'output', FileHandle)fwrite_string (FileHandle, a)
fnew_line (FileHandle)fwrite_string (FileHandle, l1)
fnew_line (FileHandle)fwrite_string (FileHandle, l2)
fnew_line (FileHandle)