码迷,mamicode.com
首页 > 编程语言 > 详细

Java OCR 图像智能字符识别技术,可识别中文

时间:2018-08-16 16:36:00      阅读:167      评论:0      收藏:0      [点我收藏+]

标签:sim   nts   creat   sel   readers   清晰度   lse   you   tar   

http://blog.csdn.net/zhoushuyan/article/details/5948289

验证码的OCR方式识别

http://ykf.iteye.com/blog/212431

 

几天一直在研究OCR技术,据我了解的情况,国内最专业的OCR软件只有2家,清华TH-OCR和汉王OCR,看了很多的OCR技术发现好多对英文与数字的支持都很好,可惜很多都不支持中文字符。Asprise-OCR,Tesseract 3.0以前的版本,都不支持中文,其实我用了下Asprise-OCR算是速度比较的快了,可惜他鄙视中文,这个没有办法,正好这段时间知名的开源OCR引擎Tesseract 3.0版本发布了,他给我们带来的好消息就是支持中文,相关的下载项目网站是:http://code.google.com/p/tesseract-ocr

虽然速度不是很客观可是毕竟人家开始支持中文也算是不错的,一个英文的语言包大概是1.8M,中文简体的语言包是39.5M,中文繁体的语言包是53M,这样就知道为什么识别中文慢的原因了

package com.ocr;

 

import java.awt.Graphics2D;

 

import java.awt.color.ColorSpace;

import java.awt.geom.AffineTransform;

import java.awt.image.AffineTransformOp;

import java.awt.image.BufferedImage;

import java.awt.image.ColorConvertOp;

import java.awt.image.ColorModel;

import java.awt.image.MemoryImageSource;

import java.awt.image.PixelGrabber;

 

 

/**

 *

 * 图像过滤,增强OCR识别成功率

 *

 */

public class ImageFilter {

    private BufferedImage image;

 

    private int iw, ih;

 

    private int[] pixels;

 

    public ImageFilter(BufferedImage image) {

       this.image = image;

       iw = image.getWidth();

       ih = image.getHeight();

       pixels = new int[iw * ih];

    }

 

    /** 图像二值化 */

    public BufferedImage changeGrey() {

       PixelGrabber pg = new PixelGrabber(image.getSource(), 0, 0, iw, ih, pixels, 0,iw);

       try {

           pg.grabPixels();

       } catch (InterruptedException e) {

           e.printStackTrace();

       }

       // 设定二值化的域值,默认值为100

       int grey = 100;

       // 对图像进行二值化处理,Alpha值保持不变

       ColorModel cm = ColorModel.getRGBdefault();

       for (int i = 0; i < iw * ih; i++) {

           int red, green, blue;

           int alpha = cm.getAlpha(pixels[i]);

           if (cm.getRed(pixels[i]) > grey) {

              red = 255;

           } else {

              red = 0;

           }

 

           if (cm.getGreen(pixels[i]) > grey) {

              green = 255;

           } else {

              green = 0;

           }

 

           if (cm.getBlue(pixels[i]) > grey) {

              blue = 255;

           } else {

              blue = 0;

           }

 

           pixels[i] = alpha << 24 | red << 16 | green << 8 | blue;

       }

       // 将数组中的象素产生一个图像

       return ImageIOHelper.imageProducerToBufferedImage(new MemoryImageSource(iw, ih,pixels, 0, iw));

    }

 

    /** 提升清晰度,进行锐化 */

    public BufferedImage sharp() {

       PixelGrabber pg = new PixelGrabber(image.getSource(), 0, 0, iw, ih, pixels, 0,iw);

       try {

           pg.grabPixels();

       } catch (InterruptedException e) {

           e.printStackTrace();

       }

 

       // 象素的中间变量

       int tempPixels[] = new int[iw * ih];

       for (int i = 0; i < iw * ih; i++) {

           tempPixels[i] = pixels[i];

       }

       // 对图像进行尖锐化处理,Alpha值保持不变

       ColorModel cm = ColorModel.getRGBdefault();

       for (int i = 1; i < ih - 1; i++) {

           for (int j = 1; j < iw - 1; j++) {

              int alpha = cm.getAlpha(pixels[i * iw + j]);

 

              // 对图像进行尖锐化

              int red6 = cm.getRed(pixels[i * iw + j + 1]);

              int red5 = cm.getRed(pixels[i * iw + j]);

              int red8 = cm.getRed(pixels[(i + 1) * iw + j]);

              int sharpRed = Math.abs(red6 - red5) + Math.abs(red8 - red5);

 

              int green5 = cm.getGreen(pixels[i * iw + j]);

              int green6 = cm.getGreen(pixels[i * iw + j + 1]);

              int green8 = cm.getGreen(pixels[(i + 1) * iw + j]);

              int sharpGreen = Math.abs(green6 - green5) + Math.abs(green8 - green5);

 

              int blue5 = cm.getBlue(pixels[i * iw + j]);

              int blue6 = cm.getBlue(pixels[i * iw + j + 1]);

              int blue8 = cm.getBlue(pixels[(i + 1) * iw + j]);

              int sharpBlue = Math.abs(blue6 - blue5) + Math.abs(blue8 - blue5);

 

              if (sharpRed > 255) {

                  sharpRed = 255;

              }

              if (sharpGreen > 255) {

                  sharpGreen = 255;

              }

              if (sharpBlue > 255) {

                  sharpBlue = 255;

              }

 

              tempPixels[i * iw + j] = alpha << 24 | sharpRed << 16 | sharpGreen << 8 | sharpBlue;

           }

       }

 

       // 将数组中的象素产生一个图像

       return ImageIOHelper.imageProducerToBufferedImage(new MemoryImageSource(iw, ih, tempPixels, 0, iw));

    }

 

    /** 中值滤波 */

    public BufferedImage median() {

       PixelGrabber pg = new PixelGrabber(image.getSource(), 0, 0, iw, ih, pixels, 0,iw);

       try {

           pg.grabPixels();

       } catch (InterruptedException e) {

           e.printStackTrace();

       }

       // 对图像进行中值滤波,Alpha值保持不变

       ColorModel cm = ColorModel.getRGBdefault();

       for (int i = 1; i < ih - 1; i++) {

           for (int j = 1; j < iw - 1; j++) {

              int red, green, blue;

              int alpha = cm.getAlpha(pixels[i * iw + j]);

 

              // int red2 = cm.getRed(pixels[(i - 1) * iw + j]);

              int red4 = cm.getRed(pixels[i * iw + j - 1]);

              int red5 = cm.getRed(pixels[i * iw + j]);

              int red6 = cm.getRed(pixels[i * iw + j + 1]);

              // int red8 = cm.getRed(pixels[(i + 1) * iw + j]);

 

              // 水平方向进行中值滤波

              if (red4 >= red5) {

                  if (red5 >= red6) {

                     red = red5;

                  } else {

                     if (red4 >= red6) {

                         red = red6;

                     } else {

                         red = red4;

                     }

                  }

              } else {

                  if (red4 > red6) {

                     red = red4;

                  } else {

                      if (red5 > red6) {

                         red = red6;

                     } else {

                         red = red5;

                     }

                  }

              }

 

              // int green2 = cm.getGreen(pixels[(i - 1) * iw + j]);

              int green4 = cm.getGreen(pixels[i * iw + j - 1]);

              int green5 = cm.getGreen(pixels[i * iw + j]);

              int green6 = cm.getGreen(pixels[i * iw + j + 1]);

              // int green8 = cm.getGreen(pixels[(i + 1) * iw + j]);

 

              // 水平方向进行中值滤波

              if (green4 >= green5) {

                  if (green5 >= green6) {

                     green = green5;

                  } else {

                     if (green4 >= green6) {

                         green = green6;

                     } else {

                         green = green4;

                     }

                  }

              } else {

                  if (green4 > green6) {

                      green = green4;

                  } else {

                     if (green5 > green6) {

                         green = green6;

                     } else {

                         green = green5;

                     }

                  }

              }

 

              // int blue2 = cm.getBlue(pixels[(i - 1) * iw + j]);

              int blue4 = cm.getBlue(pixels[i * iw + j - 1]);

              int blue5 = cm.getBlue(pixels[i * iw + j]);

              int blue6 = cm.getBlue(pixels[i * iw + j + 1]);

              // int blue8 = cm.getBlue(pixels[(i + 1) * iw + j]);

 

              // 水平方向进行中值滤波

              if (blue4 >= blue5) {

                  if (blue5 >= blue6) {

                     blue = blue5;

                  } else {

                     if (blue4 >= blue6) {

                         blue = blue6;

                     } else {

                         blue = blue4;

                     }

                  }

              } else {

                  if (blue4 > blue6) {

                     blue = blue4;

                  } else {

                     if (blue5 > blue6) {

                         blue = blue6;

                     } else {

                         blue = blue5;

                     }

                  }

              }

              pixels[i * iw + j] = alpha << 24 | red << 16 | green << 8 | blue;

           }

       }

 

       // 将数组中的象素产生一个图像

       return ImageIOHelper.imageProducerToBufferedImage(new MemoryImageSource(iw, ih,pixels, 0, iw));

    }

 

    /** 线性灰度变换 */

    public BufferedImage lineGrey() {

       PixelGrabber pg = new PixelGrabber(image.getSource(), 0, 0, iw, ih, pixels, 0,iw);

       try {

           pg.grabPixels();

       } catch (InterruptedException e) {

           e.printStackTrace();

       }

       // 对图像进行进行线性拉伸,Alpha值保持不变

       ColorModel cm = ColorModel.getRGBdefault();

       for (int i = 0; i < iw * ih; i++) {

           int alpha = cm.getAlpha(pixels[i]);

           int red = cm.getRed(pixels[i]);

           int green = cm.getGreen(pixels[i]);

           int blue = cm.getBlue(pixels[i]);

 

           // 增加了图像的亮度

           red = (int) (1.1 * red + 30);

           green = (int) (1.1 * green + 30);

           blue = (int) (1.1 * blue + 30);

           if (red >= 255) {

              red = 255;

           }

           if (green >= 255) {

              green = 255;

           }

           if (blue >= 255) {

              blue = 255;

           }

           pixels[i] = alpha << 24 | red << 16 | green << 8 | blue;

       }

 

       // 将数组中的象素产生一个图像

 

       return ImageIOHelper.imageProducerToBufferedImage(new MemoryImageSource(iw, ih,pixels, 0, iw));

    }

 

    /** 转换为黑白灰度图 */

    public BufferedImage grayFilter() {

       ColorSpace cs = ColorSpace.getInstance(ColorSpace.CS_GRAY);

       ColorConvertOp op = new ColorConvertOp(cs, null);

       return op.filter(image, null);

    }

 

    /** 平滑缩放 */

    public BufferedImage scaling(double s) {

       AffineTransform tx = new AffineTransform();

       tx.scale(s, s);

       AffineTransformOp op = new AffineTransformOp(tx, AffineTransformOp.TYPE_BILINEAR);

       return op.filter(image, null);

    }

 

    public BufferedImage scale(Float s) {

       int srcW = image.getWidth();

       int srcH = image.getHeight();

       int newW = Math.round(srcW * s);

       int newH = Math.round(srcH * s);

       // 先做水平方向上的伸缩变换

       BufferedImage tmp=new BufferedImage(newW, newH, image.getType());

       Graphics2D g= tmp.createGraphics();

       for (int x = 0; x < newW; x++) {

           g.setClip(x, 0, 1, srcH);

           // 按比例放缩

           g.drawImage(image, x - x * srcW / newW, 0, null);

       }

 

        // 再做垂直方向上的伸缩变换

       BufferedImage dst = new BufferedImage(newW, newH, image.getType());

       g = dst.createGraphics();

       for (int y = 0; y < newH; y++) {

           g.setClip(0, y, newW, 1);

           // 按比例放缩

           g.drawImage(tmp, 0, y - y * srcH / newH, null);

       }

       return dst;

    }

 

}

 

 

 

 

 

 

 

package com.ocr;

 

import java.awt.Graphics2D;

 

import java.awt.Image;

import java.awt.Toolkit;

import java.awt.image.BufferedImage;

import java.awt.image.DataBufferByte;

import java.awt.image.ImageProducer;

import java.awt.image.WritableRaster;

import java.io.File;

import java.io.IOException;

import java.util.Iterator;

import java.util.Locale;

 

import javax.imageio.IIOImage;

import javax.imageio.ImageIO;

import javax.imageio.ImageReader;

import javax.imageio.ImageWriteParam;

import javax.imageio.ImageWriter;

import javax.imageio.metadata.IIOMetadata;

import javax.imageio.stream.ImageInputStream;

import javax.imageio.stream.ImageOutputStream;

import javax.swing.JOptionPane;

 

 

 

import com.sun.media.imageio.plugins.tiff.TIFFImageWriteParam;

 

public class ImageIOHelper {

    public ImageIOHelper() {

    }

 

    public static File createImage(File imageFile, String imageFormat) {

       File tempFile = null;

       try {

           Iterator<ImageReader> readers = ImageIO.getImageReadersByFormatName(imageFormat);

           ImageReader reader = readers.next();

 

           ImageInputStream iis = ImageIO.createImageInputStream(imageFile);

           reader.setInput(iis);

           // Read the stream metadata

           IIOMetadata streamMetadata = reader.getStreamMetadata();

 

           // Set up the writeParam

           TIFFImageWriteParam tiffWriteParam = new TIFFImageWriteParam(Locale.US);

           tiffWriteParam.setCompressionMode(ImageWriteParam.MODE_DISABLED);

 

           // Get tif writer and set output to file

           Iterator<ImageWriter> writers = ImageIO.getImageWritersByFormatName("tiff");

           ImageWriter writer = writers.next();

 

           BufferedImage bi = reader.read(0);

           IIOImage image = new IIOImage(bi, null, reader.getImageMetadata(0));

           tempFile = tempImageFile(imageFile);

           ImageOutputStream ios = ImageIO.createImageOutputStream(tempFile);

           writer.setOutput(ios);

           writer.write(streamMetadata, image, tiffWriteParam);

           ios.close();

 

           writer.dispose();

           reader.dispose();

       } catch (Exception exc) {

           exc.printStackTrace();

       }

       return tempFile;

    }

 

    public static File createImage(BufferedImage bi) {

       File tempFile = null;

       try {

           tempFile = File.createTempFile("tempImageFile", ".tif");

           tempFile.deleteOnExit();

           TIFFImageWriteParam tiffWriteParam = new TIFFImageWriteParam(Locale.US);

           tiffWriteParam.setCompressionMode(ImageWriteParam.MODE_DISABLED);

 

           // Get tif writer and set output to file

           Iterator<ImageWriter> writers = ImageIO.getImageWritersByFormatName("tiff");

           ImageWriter writer = writers.next();

 

           IIOImage image = new IIOImage(bi, nullnull);

           tempFile = tempImageFile(tempFile);

           ImageOutputStream ios = ImageIO.createImageOutputStream(tempFile);

           writer.setOutput(ios);

           writer.write(null, image, tiffWriteParam);

           ios.close();

           writer.dispose();

       } catch (Exception exc) {

           exc.printStackTrace();

       }

       return tempFile;

    }

 

    public static File tempImageFile(File imageFile) {

       String path = imageFile.getPath();

       StringBuffer strB = new StringBuffer(path);

       strB.insert(path.lastIndexOf(‘.‘), 0);

       return new File(strB.toString().replaceFirst("(?<=//.)(//w+)$", "tif"));

    }

 

    public static BufferedImage getImage(File imageFile) {

       BufferedImage al = null;

       try {

           String imageFileName = imageFile.getName();

           String imageFormat = imageFileName.substring(imageFileName.lastIndexOf(‘.‘) + 1);

           Iterator<ImageReader> readers = ImageIO.getImageReadersByFormatName(imageFormat);

           ImageReader reader = readers.next();

 

           if (reader == null) {

              JOptionPane.showConfirmDialog(null,

                     "Need to install JAI Image I/O package./nhttps://jai-imageio.dev.java.net");

              return null;

           }

 

           ImageInputStream iis = ImageIO.createImageInputStream(imageFile);

           reader.setInput(iis);

 

           al = reader.read(0);

 

           reader.dispose();

       } catch (IOException ioe) {

           System.err.println(ioe.getMessage());

       } catch (Exception e) {

           System.err.println(e.getMessage());

       }

 

       return al;

    }

 

    public static BufferedImage imageToBufferedImage(Image image) {

       BufferedImage bufferedImage = new BufferedImage(image.getWidth(null), image.getHeight(null),

              BufferedImage.TYPE_INT_RGB);

       Graphics2D g = bufferedImage.createGraphics();

       g.drawImage(image, 0, 0, null);

       return bufferedImage;

    }

 

    public static BufferedImage imageProducerToBufferedImage(ImageProducer imageProducer) {

       returnimageToBufferedImage(Toolkit.getDefaultToolkit().createImage(imageProducer));

    }

 

    public static byte[] image_byte_data(BufferedImage image) {

       WritableRaster raster = image.getRaster();

       DataBufferByte buffer = (DataBufferByte) raster.getDataBuffer();

       return buffer.getData();

    }

}

 

 

 

 

 

 

package com.ocr;

 

import java.io.BufferedReader;

 

 

import java.io.File;

import java.io.FileInputStream;

import java.io.InputStreamReader;

import java.util.ArrayList;

import java.util.List;

 

import org.jdesktop.swingx.util.OS;

 

public class OCR {

    private final String LANG_OPTION = "-l";

    private final String EOL = System.getProperty("line.separator");

    private String tessPath = new File("tesseract").getAbsolutePath();

    //private String tessPath="C://Program Files (x86)//Tesseract-OCR//";

    public String recognizeText(File imageFile, String imageFormat) throws Exception {

       File tempImage = ImageIOHelper.createImage(imageFile, imageFormat);

       File outputFile = new File(imageFile.getParentFile(), "output");

       StringBuffer strB = new StringBuffer();

       List<String> cmd = new ArrayList<String>();

       if (OS.isWindowsXP()) {

           cmd.add(tessPath + "//tesseract");

           //cmd.add(tessPath + "//Tesseract-OCR");

       } else if (OS.isLinux()) {

           cmd.add("tesseract");

       } else {

           //cmd.add(tessPath + "//Tesseract-OCR");

           cmd.add(tessPath + "//tesseract");

       }

           cmd.add(""); 

            cmd.add(outputFile.getName()); 

            cmd.add(LANG_OPTION); 

            cmd.add("chi_sim");

            cmd.add("eng"); 

 

       ProcessBuilder pb = new ProcessBuilder();

       pb.directory(imageFile.getParentFile());

 

       cmd.set(1, tempImage.getName());

       pb.command(cmd);

       pb.redirectErrorStream(true);

       Process process = pb.start();

       //tesseract.exe 1.jpg 1 -l chi_sim

       int w = process.waitFor();

 

       // delete temp working files

       tempImage.delete();

 

       if (w == 0) {

           BufferedReader in = new BufferedReader(new InputStreamReader(newFileInputStream(outputFile

                  .getAbsolutePath()

                  + ".txt"), "UTF-8"));

 

           String str;

 

           while ((str = in.readLine()) != null) {

              strB.append(str).append(EOL);

           }

           in.close();

       } else {

           String msg;

           switch (w) {

           case 1:

              msg = "Errors accessing files. There may be spaces in your image‘s filename.";

              break;

           case 29:

              msg = "Cannot recognize the image or its selected region.";

              break;

           case 31:

              msg = "Unsupported image format.";

              break;

           default:

              msg = "Errors occurred.";

           }

           tempImage.delete();

           throw new RuntimeException(msg);

       }

       new File(outputFile.getAbsolutePath() + ".txt").delete();

       return strB.toString();

    }

}

 

 

 

 

package com.ocr;

 

import java.io.File;

 

public class Test {

 

    /**

     * @param args

     */

    public static void main(String[] args) {

       // TODO Auto-generated method stub

       OCR ocr=new OCR();

        try {

           String maybe = new OCR().recognizeText(new  File("E://temp//222.jpg"),"jpg");

           System.out.println(maybe);

       } catch (Exception e) {

           // TODO Auto-generated catch block

           e.printStackTrace();

       }

    }

 

}

 

 

由于可以第三方包加起来有点大,告诉大家一个网站 www.findjar.com去里面找你想要的包吧,需要相关包的留下邮件吧

 

 

 

java 目录结构如上图

 

效果图:

 

 

解析出来的效果

 

 

Java OCR 图像智能字符识别技术,可识别中文

标签:sim   nts   creat   sel   readers   清晰度   lse   you   tar   

原文地址:https://www.cnblogs.com/pejsidney/p/9487888.html

(0)
(0)
   
举报
评论 一句话评论(0
登录后才能评论!
分享档案
周排行
mamicode.com排行更多图片
© 2014 mamicode.com 版权所有  联系我们:gaon5@hotmail.com
迷上了代码!