I am working on OCR to recognised passport details, Since I am using Tesseract Java API. To achieve better accuracy I need to divide the whole image (can be of .png,.jpeg, .tiff) only into text regions. Is there any open source java library which separates text regions from image. Please give me any suggestions on it.
可以将文章内容翻译成中文,广告屏蔽插件可能会导致该功能失效(如失效,请关闭广告屏蔽插件后再试):
由 翻译强力驱动
问题:
回答1:
Marvin provides a method exactly for this purpose.
public static java.util.List<MarvinSegment> findTextRegions(MarvinImage imageIn, int maxWhiteSpace, int maxFontLineWidth, int minTextWidth, int grayScaleThreshold)
Input image:
Output image:
Source code:
import static marvin.MarvinPluginCollection.*; public class TextRegions{ public static void main(String[] args) { MarvinImage image = MarvinImageIO.loadImage("./res/passport.png"); MarvinImage originalImage = image.clone(); List<MarvinSegment> segments = findTextRegions(image, 15, 8, 30, 150); for(MarvinSegment s:segments){ if(s.height >= 5){ originalImage.drawRect(s.x1, s.y1, s.x2-s.x1, s.y2-s.y1, Color.red); } } MarvinImageIO.saveImage(originalImage, "./res/passport_2.png"); } }
回答2:
Your best bet is to use OpenCV (there are bindings for Java).
The problem is hard and there's no solution that works in all cases. I would check suggestions from threads like this one and try to find the best solution for your specific case.