還是車牌識別的例子
read_image (Image, 'C:/Users/LiZhiqiang/Desktop/車牌2.jpg')//讀取圖像 rgb1_to_gray (Image, GrayImage)//轉化成灰度圖像 threshold (GrayImage, Regions, 21, 163)//閾值選擇 connection (Regions, ConnectedRegions)//斷開區域 select_shape (ConnectedRegions, SelectedRegions1, 'area', 'and', 299.63, 1000)//特征選擇，選中車牌 union1 (SelectedRegions1, RegionUnion)//把車牌合並成一個區域
如下圖
shape_trans (RegionUnion, RegionTrans, 'convex')//將形狀轉換成矩形
orientation_region (RegionTrans, Phi)//得到區域的偏轉角度 area_center (RegionTrans, Area, Row, Column)//計算區域面積和中心點坐標 vector_angle_to_rigid (Row, Column, Phi, Row, Column, 0, HomMat2D)//得到擺正圖片的仿射矩陣 affine_trans_image (GrayImage, ImageAffineTrans, HomMat2D, 'constant', 'false')//仿射矩陣應用於圖片
affine_trans_region (RegionTrans, RegionAffineTrans, HomMat2D, 'nearest_neighbor')//仿射矩陣應用於區域 reduce_domain (ImageAffineTrans, RegionAffineTrans, ImageReduced)//從圖片中截取該矩形區域
后面的部分就跟上一節的車牌識別思路一樣了
threshold (ImageReduced, Regions1, 0, 152)//設定閾值 connection (Regions1, ConnectedRegions1)//斷開區域 sort_region (ConnectedRegions1, SortedRegions, 'character', 'true', 'row') read_ocr_class_mlp ('Document_0-9A-Z_NoRej.omc', OCRHandle) do_ocr_multi_class_mlp (SortedRegions, ImageReduced, OCRHandle, Class, Confidence)
