Java Tess4J Demo

Java Tess4J 实现OCR 可运行Demo

/**
 * 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
----------- ---------  ---------- -----  ----

     文件        659  2014-09-29 10:54  Java_OCR\.classpath

     文件        384  2014-09-29 09:59  Java_OCR\.project

     文件        598  2014-09-29 09:59  Java_OCR\.settings\org.eclipse.jdt.core.prefs

     文件        676  2014-09-29 10:54  Java_OCR\bin\com\recognition\software\jdeskew\ImageDeskew$HoughLine.class

     文件       3616  2014-09-29 10:54  Java_OCR\bin\com\recognition\software\jdeskew\ImageDeskew.class

     文件       1432  2014-09-29 10:54  Java_OCR\bin\com\recognition\software\jdeskew\ImageUtil.class

     文件      17021  2014-09-29 11:12  Java_OCR\bin\com\ricky\java\ocr\TessAPITest$TessDllAPIImpl.class

     文件      17810  2014-09-29 11:12  Java_OCR\bin\com\ricky\java\ocr\TessAPITest.class

     文件       4206  2014-09-29 11:12  Java_OCR\bin\com\ricky\java\ocr\Tesseract1Test$Tess1Extension.class

     文件       1520  2014-09-29 11:12  Java_OCR\bin\com\ricky\java\ocr\Tesseract1Test$Word.class

     文件       5657  2014-09-29 11:12  Java_OCR\bin\com\ricky\java\ocr\Tesseract1Test.class

     文件   14789120  2014-09-29 11:11  Java_OCR\bin\gsdll64.dll

     文件    2120704  2014-09-29 10:34  Java_OCR\bin\liblept168.dll

     文件    2000384  2014-09-29 10:34  Java_OCR\bin\libtesseract302.dll

     文件       1334  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\ITesseract.class

     文件        552  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$ETEXT_DESC.class

     文件        555  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$TessbaseAPI.class

     文件        579  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$TessMutableIterator.class

     文件        408  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$TessOcrEngineMode.class

     文件        418  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$TessOrientation.class

     文件        570  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$TessPageIterator.class

     文件        418  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$TessPageIteratorLevel.class

     文件        782  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$TessPageSegMode.class

     文件        838  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$TessPolyBlockType.class

     文件        576  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$TessResultIterator.class

     文件        399  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$TessTextlineOrder.class

     文件        414  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI$TessWritingDirection.class

     文件       8241  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI.class

     文件        556  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI1$ETEXT_DESC.class

     文件        559  2014-09-29 10:54  Java_OCR\bin\net\sourceforge\tess4j\TessAPI1$TessbaseAPI.class

............此处省略86个文件信息