Tess4J-1.2-src包

Java使用Tess4J 进行图片文字识别

/**
 * JDeskew
 */
package com.recognition.software.jdeskew;

import java.awt.image.BufferedImage;

public class ImageDeskew {

    /**
     * Representation of a line in the image.
     */
    public class HoughLine {

        // count of points in the line
        public int count = 0;
        // index in matrix.
        public int index = 0;
        // the line is represented as all x y that solve y * cos（alpha） - x *
        // sin（alpha） = d
        public double alpha;
        public double d;
    }
    // the source image
    private BufferedImage cImage;
    // the range of angles to search for lines
    private double cAlphaStart = -20;
    private double cAlphaStep = 0.2;
    private int cSteps = 40 * 5;
    // pre-calculation of sin and cos
    private double[] cSinA;
    private double[] cCosA;
    // range of d
    private double cDMin;
    private double cDStep = 1.0;
    private int cDCount;
    // count of points that fit in a line
    private int[] cHMatrix;

    // constructor
    public ImageDeskew（BufferedImage image） {
        this.cImage = image;
    }

    // calculate the skew angle of the image cImage
    public double getSkewAngle（） {
        ImageDeskew.HoughLine[] hl;
        double sum = 0.0;
        int count = 0;

        // perform Hough Transformation
        calc（）;
        // top 20 of the detected lines in the image
        hl = getTop（20）;

        if （hl.length >= 20） {
            // average angle of the lines
            for （int i = 0; i < 19; i++） {
                sum += hl[i].alpha;
                count++;
            }
            return （sum / count）;
        } else {
            return 0.0d;
        }
    }

    // calculate the count lines in the image with most points
    private ImageDeskew.HoughLine[] getTop（int count） {

        ImageDeskew.HoughLine[] hl = new ImageDeskew.HoughLine[count];
        for （int i = 0; i < count; i++） {
            hl[i] = new ImageDeskew.HoughLine（）;
        }

        ImageDeskew.HoughLine tmp;

        for （int i = 0; i < （this.cHMatrix.length - 1）; i++） {
            if （this.cHMatrix[i] > hl[count - 1].count） {
                hl[count - 1].count = this.cHMatrix[i];
                hl[count - 1].index = i;
                int j = count - 1;
                while （（j > 0） && （hl[j].count > hl[j - 1].count）） {
                    tmp = hl[j];
                    hl[j] = hl[j - 1];
                    hl[j - 1] = tmp;
                    j--;
                }
            }
        }

        int alphaIndex;
        int dIndex;
        
        for （int i = 0; i < count; i++） {
            dIndex = hl[i].index / cSteps; // integer division no
            // remainder
            alphaIndex = hl[i].index - dIndex * cSteps;
            hl[i].alpha = getAlpha（alphaIndex）;
            hl[i].d = dIndex + cDMin;
        }

        re

 属性            大小     日期    时间   名称
----------- ---------  ---------- -----  ----
     目录           0  2013-09-22 11:07  Tess4J\
     目录           0  2013-09-22 11:07  Tess4J\dist\
     目录           0  2013-09-21 08:34  Tess4J\lib\
     目录           0  2013-09-21 08:35  Tess4J\nbproject\
     目录           0  2013-02-28 22:17  Tess4J\nbproject\private\
     目录           0  2013-02-28 22:17  Tess4J\src\
     目录           0  2013-02-28 22:17  Tess4J\src\com\
     目录           0  2013-02-28 22:17  Tess4J\src\com\recognition\
     目录           0  2013-02-28 22:17  Tess4J\src\com\recognition\software\
     目录           0  2013-02-28 22:17  Tess4J\src\com\recognition\software\jdeskew\
     目录           0  2013-02-28 22:17  Tess4J\src\net\
     目录           0  2013-09-21 12:06  Tess4J\src\net\sourceforge\
     目录           0  2013-09-21 12:06  Tess4J\src\net\sourceforge\tess4j\
     目录           0  2013-09-21 12:06  Tess4J\src\net\sourceforge\vietocr\
     目录           0  2013-09-21 12:12  Tess4J\tessdata\
     目录           0  2013-02-28 22:17  Tess4J\tessdata\configs\
     目录           0  2013-02-28 22:17  Tess4J\test\
     目录           0  2013-02-28 22:17  Tess4J\test\net\
     目录           0  2013-09-21 08:52  Tess4J\test\net\sourceforge\
     目录           0  2013-09-21 08:52  Tess4J\test\net\sourceforge\tess4j\
     目录           0  2013-09-21 08:52  Tess4J\test\net\sourceforge\tess4j\utiltities\
     文件        5153  2013-08-31 10:01  Tess4J\build.xml
     文件        1322  2013-09-22 11:07  Tess4J\dist\README.TXT
     文件       69037  2013-09-22 11:07  Tess4J\dist\tess4j.jar
     文件      102464  2013-02-28 22:17  Tess4J\eurotext.bmp
     文件       20359  2013-02-28 22:17  Tess4J\eurotext.gif
     文件       13065  2013-02-28 22:17  Tess4J\eurotext.pdf
     文件       14854  2013-02-28 22:17  Tess4J\eurotext.png
     文件      102598  2013-02-28 22:17  Tess4J\eurotext.tif
     文件      204383  2013-02-28 22:17  Tess4J\eurotext_deskew.png
     文件       31679  2013-02-28 22:17  Tess4J\lib\ghost4j-0.3.1.jar
............此处省略35个文件信息