因为腾讯的文字识别还是会有错别字,需要手动校验,时间和精力都不允许,所以就萌发了直接展示原文字图片给用户, 为了适配不同宽度的终端, 增加了文字图片重新排版的功能, 大部分情况表现良好, 还是会有一些bug, 后续有空再调整吧.
引入jar包
<dependency>
<groupId>net.coobird</groupId>
<artifactId>thumbnailator</artifactId>
<version>0.4.12</version>
</dependency>
代码
使用说明:
- 1.通过binaryPic()方法, 将手机拍摄的文字图片进行二值化
- 2.将上面的二值化图片作为modifyPicWidth()方法的参数, 并设置新生成图片的宽度, 执行就可以了
- 3.重新排版的原理就是先切割成单个文字图片, 再重新拼接, 这个过程中需要记录换行的位置.
- 4.存在的问题, 对于标点符号, 会出现忽上忽下的问题, 暂时没有什么好办法解决
package com.example.springboot2.picture;
import net.coobird.thumbnailator.Thumbnails;
import org.junit.jupiter.api.Test;
import javax.imageio.ImageIO;
import java.awt.image.BufferedImage;
import java.io.File;
import java.io.IOException;
import java.util.*;
public class Test0315 {
// 横向切分块数
private static final int widthSplitNum = 12;
// 纵向切分块数
private static final int heightSplitNum = 14;
// 面积小于50像素的字符,舍弃掉
private static final int ignoreArea = 30;
// 首行缩进宽度, 超过该值就换行
private static final int textIndent = 80;
// 行尾空白宽度, 超过该值就换行
private static final int rowBlankWidth = 75;
/**
* 二值化图片
* @throws IOException
*/
@Test
public void binaryPic() throws IOException {
long start = System.currentTimeMillis();
// 黑白图片
String fromPic = "d:\\IMG_20221001_064715.jpg";
File file = new File(fromPic);
BufferedImage bufferedImage = ImageIO.read(file);
// 压缩图片 275ms
BufferedImage compactImage = Thumbnails.of(bufferedImage).size(2000, 2000).asBufferedImage();
// 切割图片,分别处理,防止出现大片黑斑 1ms
Map<Integer, BufferedImage> map = splitImage2(compactImage);
long startTime = System.currentTimeMillis();
// 二值化
for (Integer key : map.keySet()) {
BufferedImage binaryImage = binaryImage(map.get(key));
map.put(key, binaryImage);
}
System.out.println("1:" + (System.currentTimeMillis() - startTime));
// 合并图片
BufferedImage destImage = mergeImage(map);
System.out.println("2:" + (System.currentTimeMillis() - startTime));
//System.out.println("耗时-1:" + (System.currentTimeMillis()- start) + "ms");
File newFile2 = new File("d:\\dd3.png");
ImageIO.write(destImage, "png", newFile2);
//System.out.println("耗时0:" + (System.currentTimeMillis()- start) + "ms");
}
/**
* 调整文字图片宽度
*/
@Test
public void modifyPicWidth() throws IOException {
long start = System.currentTimeMillis();
// 页面宽度
int maxWidth = 1000;
// 字符与字符之间的间隔
int colSpace = 8;
// 行与行之间的间隔
int rowSpace = 15;
// 黑白图片
String fromPic = "d:\\dd3.png";
//String fromPic = "E:\\documents\\DCIM\\unfinished\\邗江县资料本\\IMG_20230215_213145.png";
File file = new File(fromPic);
BufferedImage bufferedImage = ImageIO.read(file);
// 去除图片左右两边的污点
BufferedImage cutLRImage = cutLR(bufferedImage);
// 将字符行染黑
BufferedImage expandWhite = expandWhite2(cutLRImage);
// 根据上面确定的字符行位置,提取出字符行
List<BufferedImage> list = getCharRow4(expandWhite, cutLRImage);
List<BufferedImage> list2 = new ArrayList<>();
int i = 1;
// 切割字符
for (BufferedImage item : list) {
List<BufferedImage> charImages = splitChar3(item);
list2.addAll(charImages);
}
BufferedImage bufferedImage2 = mergeImage5(list2, maxWidth, colSpace, rowSpace);
File newFile = new File("d:\\ccc.png");
ImageIO.write(bufferedImage2, "png", newFile);
System.out.println("耗时2:" + (System.currentTimeMillis()- start) + "ms");
}
/**
* 同一水平线上的两个黑色像素点,如果他们之家的间距小于设定的阈值,那么将他们之间的所有白色像素点变成黑色像素点
* @param bufferedImage
* @return
*/
private BufferedImage expandWhite2(BufferedImage bufferedImage) {
BufferedImage grayImage = new BufferedImage(bufferedImage.getWidth(), bufferedImage.getHeight(), bufferedImage.getType());
//long start1 = System.currentTimeMillis();
grayImage.setData(bufferedImage.getData());
//System.out.println("expandWhite2耗时1:" + (System.currentTimeMillis()- start1) + "ms");
int width = bufferedImage.getWidth();
int height = bufferedImage.getHeight();
map3 = new HashMap<>();
for (int col=0; col<height; col++) {
for (int row = 0; row < width; row++) {
int color = bufferedImage.getRGB(row, col);
if ((color & 0xff) == 0) {
map3.put(col, 1);
break;
}
}
}
// 抹黑字符,确定字符行的位置,便于后续的调整行间距
// 从左往右扫描,遇到黑色像素,就将他左边的像素抹黑
int threshold = 120;
// 第一个黑色像素点下标
int start = -1;
int end = -1;
// 两个黑色像素点之间的白色像素点个数
int whiteCount = 0;
for (int j=0; j<height; j++) {
if (!map3.containsKey(j)) {
continue;
}
for (int i=0; i<width; i++) {
int color = grayImage.getRGB(i, j);
int red = color & 0xFF;
// 黑色像素点
if (red == 0) {
if (start == -1) {
start = i;
} else {
if (whiteCount == 0) {
start = i;
} else {
if (whiteCount < threshold) {
for (int k=start+1; k<i; k++) {
grayImage.setRGB(k, j, 0);
}
}
start = i;
whiteCount = 0;
}
}
} else if (red == 255 && start != -1) {
// 白色像素点
whiteCount++;
}
}
start = -1;
whiteCount = 0;
}
//System.out.println("expandWhite2耗时2" + (System.currentTimeMillis()- start1) + "ms");
return grayImage;
}
/**
* 处理整个图片
* 截图,去除左右边框
* @param bufferedImage
* @return
*/
private BufferedImage cutLR(BufferedImage bufferedImage) {
//long start = System.currentTimeMillis();
// 横坐标上的映射图片,如果首行连续空白像素的个数超过阈值,就认为连续空白像素前面的都是污点,需要清除掉
int threshold = 10;
int ignoreHeight = 5;
int width = bufferedImage.getWidth();
int height = bufferedImage.getHeight();
// 存储黑色行号,key为x轴坐标,value为对应x坐标上的黑色像素个数
Map<Integer, Integer> map = new HashMap<>();
// 统计
for (int i = 0; i < width; i++) {
int blackCount = 0;
for (int j = 0; j < height; j++) {
// getRGB()方法,根据手册,其返回的int型数据(32位)为ARGB格式,其中ARGB各占8bit
int color = bufferedImage.getRGB(i, j);
int b = color & 0xff;
if (b == 0) {
blackCount++;
if (blackCount > ignoreHeight) {
break;
}
}
}
map.put(i, blackCount);
}
//System.out.println("耗时-4:" + (System.currentTimeMillis()- start) + "ms");
int whiteCount = 0;
int startIndex = -1;
int endIndex = width-1;
// 找出起始像素点
for (int i = 0; i < width/2; i++) {
int value = map.get(i);
// x轴上像素个数小于3的都认为是空白
if (value < ignoreHeight) {
whiteCount++;
if (whiteCount >= threshold) {
startIndex = i;
}
} else {
whiteCount = 0;
}
}
// 找出结束像素点
whiteCount = 0;
for (int i = width-1; i > width/2; i--) {
int value = map.get(i);
if (value < ignoreHeight) {
whiteCount++;
if (whiteCount >= threshold) {
endIndex = i;
}
} else {
whiteCount = 0;
}
}
// 截掉图片左右两个空白部分
BufferedImage grayImage = bufferedImage.getSubimage(startIndex, 0, endIndex-startIndex, height);
return grayImage;
}
/**
* 去除单个字符上下边框
* @param bufferedImage
* @return
*/
private BufferedImage cutUD(BufferedImage bufferedImage) {
int startIndex = 0;
// 找到起始像素点
out:
for (int j = 0; j < bufferedImage.getHeight(); j++) {
for (int i = 0; i < bufferedImage.getWidth(); i++) {
// getRGB()方法,根据手册,其返回的int型数据(32位)为ARGB格式,其中ARGB各占8bit
int color = bufferedImage.getRGB(i, j);
int b = color & 0xff;
if (b == 0) {
startIndex = j;
break out;
}
}
}
// 找到结束像素点
int endIndex = bufferedImage.getWidth()-1;
out:
for (int j = bufferedImage.getHeight()-1; j > -1; j--) {
for (int i = 0; i < bufferedImage.getWidth(); i++) {
int color = bufferedImage.getRGB(i, j);
int b = color & 0xff;
if (b == 0) {
endIndex = j;
break out;
}
}
}
// 截掉图片上下两个空白部分
// 不清除上下边界
if (bufferedImage.getWidth() < 30) {
return bufferedImage;
} else {
return bufferedImage.getSubimage(0, startIndex, bufferedImage.getWidth(), endIndex-startIndex+1);
}
}
/**
* 去除单个字符左右边框
* @param bufferedImage
* @return
*/
private BufferedImage cutLR2(BufferedImage bufferedImage) {
int startIndex = 0;
// 找到起始像素点
out:
for (int i = 0; i < bufferedImage.getWidth(); i++) {
for (int j = 0; j < bufferedImage.getHeight(); j++) {
// getRGB()方法,根据手册,其返回的int型数据(32位)为ARGB格式,其中ARGB各占8bit
int color = bufferedImage.getRGB(i, j);
int b = color & 0xff;
if (b == 0) {
startIndex = i;
break out;
}
}
}
// 找到结束像素点
int endIndex = bufferedImage.getWidth()-1;
out:
for (int i = bufferedImage.getWidth()-1; i > -1; i--) {
for (int j = 0; j < bufferedImage.getHeight(); j++) {
int color = bufferedImage.getRGB(i, j);
int b = color & 0xff;
if (b == 0) {
endIndex = i;
break out;
}
}
}
return bufferedImage.getSubimage(startIndex, 0, endIndex-startIndex+1, bufferedImage.getHeight());
}
/**
* 去除单个字符左边空白部分
* @param bufferedImage
* @return
*/
private BufferedImage cutL(BufferedImage bufferedImage) {
int startIndex = 0;
// 找到起始像素点
out:
for (int i = 0; i < bufferedImage.getWidth(); i++) {
for (int j = 0; j < bufferedImage.getHeight(); j++) {
// getRGB()方法,根据手册,其返回的int型数据(32位)为ARGB格式,其中ARGB各占8bit
int color = bufferedImage.getRGB(i, j);
int b = color & 0xff;
if (b == 0) {
startIndex = i;
break out;
}
}
}
return bufferedImage.getSubimage(startIndex, 0, bufferedImage.getWidth()-startIndex, bufferedImage.getHeight());
}
/**
* 去除单个字符右边空白部分
* @param bufferedImage
* @return
*/
private BufferedImage cutR(BufferedImage bufferedImage) {
// 找到结束像素点
int endIndex = bufferedImage.getWidth()-1;
out:
for (int i = bufferedImage.getWidth()-1; i > -1; i--) {
for (int j = 0; j < bufferedImage.getHeight(); j++) {
int color = bufferedImage.getRGB(i, j);
int b = color & 0xff;
if (b == 0) {
endIndex = i;
break out;
}
}
}
return bufferedImage.getSubimage(0, 0, endIndex+1, bufferedImage.getHeight());
}
/**
* 对一行字符进行分割
* 紧贴分割,且不会造成将一个正常字符分割成两个部分的情况
* 如果过度分割,那么就进行合并
* 合并的原理就是只要两个相邻的字符合并后不超过最大字符宽度,就认为是一个字符,否则就是两个字符
* @param bufferedImage
* @return
*/
private List<BufferedImage> splitChar3(BufferedImage bufferedImage) {
List<BufferedImage> charImages = new ArrayList<>();
int width = bufferedImage.getWidth();
int height = bufferedImage.getHeight();
// 横坐标上的映射图片,如果x坐标上的像素个数小于这个阈值,就认为是噪点,舍弃掉,值越大,达到分割的间距越小,分的越细
//int threshold = 2;
// 字符间距,值越小,达到分割的间距越小,分的越细
//int interval = 1;
// 横坐标上的映射图片,如果x坐标上的像素个数小于这个阈值,就认为是噪点,舍弃掉,值越大,达到分割的间距越小,分的越细
int threshold = 1;
// 字符间距,值越小,达到分割的间距越小,分的越细
int interval = 2;
// 存储黑色像素在x轴上的映射,key为x轴坐标,value为对应x坐标上的黑色像素个数
Map<Integer, Integer> map = new HashMap<>();
// 存储字符起始和结束下标,key为起始下标,value为结束下标后一位,二者差就是字符实际宽度
Map<Integer, Integer> charsMap = new LinkedHashMap<>();
// 统计
for (int i = 0; i < width; i++) {
int blackCount = 0;
for (int j = 0; j < height; j++) {
// getRGB()方法,根据手册,其返回的int型数据(32位)为ARGB格式,其中ARGB各占8bit
int color = bufferedImage.getRGB(i, j);
int b = color & 0xff;
if (b == 0) {
blackCount++;
if (blackCount > threshold) {
break;
}
}
}
map.put(i, blackCount);
}
// 从左到右遍历时,记录白色像素点的宽度,超过这个值,就表示当前处于字符之间的空白处
int whiteCount = 0;
int startIndex = -1;
// 单个字符的最小宽度,用于判断当前字符是否需要合并
// 存在“出”字被分成两个部分,左边宽度大于bufferedImage.getHeight()*2/3,右边宽度为1像素,导致不能合并,需要增大minWidth
//int minWidth = bufferedImage.getHeight()*2/3;
// 对于字符行是弯曲的情况,直接使用图片高度不适用
//int minWidth = bufferedImage.getHeight();
// 对于“小”字这种宽度大于高度的,如果他所在行的有效高度正好等于他的高度,就会造成”小“被分割成两个部分,所以需要额外增加1/20和一个像素
int maxHeight = getMaxHeight(bufferedImage);
int minWidth = maxHeight + maxHeight/20 + 1;
// 单个字符的最大宽度
// 存在“呀,”合并的情况,因为它们的宽度小于bufferedImage.getHeight()*5/4,需要缩小他的值
//int maxWidth = bufferedImage.getHeight()*5/4;
// 对于字符行是弯曲的情况,直接使用图片高度不适用
//int maxWidth = bufferedImage.getHeight();
int maxWidth = minWidth;
int lastKey = -1;
int lastValue = -1;
// 找出起始像素点
for (int i = 0; i < width; i++) {
int value = map.get(i);
// 字符起始下标
if (startIndex == -1) {
// 对于高度小于设定阈值的坐标,直接舍弃
if (value >= threshold) {
startIndex = i;
}
} else {
// x轴上像素个数小于3的都认为是空白
if (value < threshold) {
whiteCount++;
} else {
// 对于一个字符中间存在空白的情况,需要排除掉
whiteCount = 0;
}
if (whiteCount > interval) {
int charWidth = i-interval-startIndex;
// 防止一个字符被分割成两个部分
if (charWidth < minWidth) {
// 如果是类似“儿”这样的左右分开的字符
if (lastKey == -1) {
lastKey = startIndex;
lastValue = i-interval;
} else {
if (i-interval-lastKey > maxWidth) {
charsMap.put(lastKey, lastValue);
lastKey = startIndex;
lastValue = i-interval;
} else {
/*charsMap.put(lastKey, i-interval);
lastKey = -1;
lastValue = -1;*/
// 存在一个字符被分割成多个情况,这时需要多次合并
lastValue = i-interval;
}
}
} else {
if (lastKey != -1) {
charsMap.put(lastKey, lastValue);
lastKey = -1;
lastValue = -1;
}
charsMap.put(startIndex, i-interval);
}
startIndex = -1;
whiteCount = 0;
}
}
}
// 处理行尾的字符
if (lastKey != -1 && startIndex != -1) {
int i = width;
// 后一个字符的宽度
int charWidth = i-interval-startIndex;
// 防止一个字符被分割成两个部分
if (charWidth < minWidth) {
// 如果是类似“儿”这样的左右分开的字符
if (i-interval-lastKey > maxWidth) {
charsMap.put(lastKey, lastValue);
charsMap.put(startIndex, i-interval);
} else {
// 存在一个字符被分割成多个情况,这时需要多次合并
charsMap.put(lastKey, i-interval);
}
} else {
charsMap.put(lastKey, lastValue);
charsMap.put(startIndex, i-interval);
}
lastKey = -1;
lastValue = -1;
startIndex = -1;
}
if (lastKey != -1 && startIndex == -1) {
// 过滤噪点,暂时不考虑与最后一个字符合并
if (lastValue - lastKey > 1) {
charsMap.put(lastKey, lastValue);
lastKey = -1;
lastValue = -1;
}
}
if (startIndex != -1 && lastKey == -1) {
charsMap.put(startIndex, width);
startIndex = -1;
}
// 清除无效字符
Iterator<Map.Entry<Integer, Integer>> itr = charsMap.entrySet().iterator();
while (itr.hasNext()) {
Map.Entry<Integer, Integer> entry = itr.next();
Integer key = entry.getKey();
int value = entry.getValue();
if (value <= key) {
itr.remove();
continue;
}
BufferedImage grayImage = bufferedImage.getSubimage(key,0,value-key,height);
if (isInvalidChar(grayImage)) {
itr.remove();
}
}
// 是否是第一个有效字符
boolean flag = true;
// 第几个元素
int num = 1;
// 根据上面已经标注好的字符下标,提取字符图片
for (Integer key : charsMap.keySet()) {
int value = charsMap.get(key);
BufferedImage grayImage = null;
try {
grayImage = bufferedImage.getSubimage(key,0,value-key,height);
} catch (Exception e) {
e.printStackTrace();
}
// 首行缩进
if (flag && key > textIndent) {
grayImage = bufferedImage.getSubimage(key-textIndent,0,value-key+textIndent,height);
// 将缩进部分设置为纯白色
for (int row=0; row<textIndent; row++) {
for (int col=0; col<height; col++) {
grayImage.setRGB(row, col, 0xFFFFFF);
}
}
}
// 如果最后一个字符后面有空白, 就认为下一行需要换行
if ((num++) == charsMap.size() && bufferedImage.getWidth() - value > rowBlankWidth) {
flag = true;
grayImage = bufferedImage.getSubimage(key,0,value-key+rowBlankWidth,height);
// 将行末尾空白部分设置为纯白色
for (int row=value-key; row<value-key+rowBlankWidth; row++) {
for (int col=0; col<height; col++) {
grayImage.setRGB(row, col, 0xFFFFFF);
}
}
}
// 清除字符上下空白部分
BufferedImage cutUDImage = cutUD(grayImage);
// 再次进行分割
//if (cutUDImage.getWidth() > bufferedImage.getHeight() && cutUDImage.getWidth() > cutUDImage.getHeight()*2) {
// 案例:“这样”字符宽80像素,高39像素,maxWidth=42,不满足cutUDImage.getWidth() > maxWidth*1.92,没有进行分割
//if (cutUDImage.getWidth() > maxHeight*1.92) {
if (!flag && cutUDImage.getWidth() > maxHeight*splitThreshold) {
List<BufferedImage> list2 = splitChar4(cutUDImage);
charImages.addAll(list2);
} else {
charImages.add(cutUDImage);
}
flag = false;
}
// “个儿”这两个字符会分成三个部分,需要将后面两个部分合在一起
for (int i=0; i<charImages.size()-1; i++) {
BufferedImage bufferedImage1 = charImages.get(i);
if (bufferedImage1.getWidth() * 2 < bufferedImage1.getHeight()
&& bufferedImage1.getWidth() * 3 > bufferedImage1.getHeight()) {
BufferedImage bufferedImage2 = charImages.get(i+1);
// 两个字符的宽度和高度都差不多,且高度都大致等于宽度的两倍
if (bufferedImage2.getWidth() * 1.8 < bufferedImage2.getHeight()
&& bufferedImage2.getWidth() * 3 > bufferedImage2.getHeight()
&& Math.abs(bufferedImage1.getHeight()-bufferedImage2.getHeight()) < 5) {
// 两个字符合并后中间的空白宽度
int blankWidth = 4;
int newWidth = bufferedImage1.getWidth() + blankWidth + bufferedImage2.getWidth();
int newHeight = Math.max(bufferedImage1.getHeight(), bufferedImage2.getHeight());
// 进行合并
BufferedImage grayImage2 = new BufferedImage(newWidth, newHeight, bufferedImage.getType());
// 设置白色背景
for (int row=0; row<newWidth; row++) {
for (int col=0; col<newHeight; col++) {
grayImage2.setRGB(row, col, 0xFFFFFF);
}
}
for (int row=0; row<bufferedImage1.getWidth(); row++) {
for (int col=0; col<bufferedImage1.getHeight(); col++) {
grayImage2.setRGB(row, col, bufferedImage1.getRGB(row, col));
}
}
for (int row=0; row<bufferedImage2.getWidth(); row++) {
for (int col=0; col<bufferedImage2.getHeight(); col++) {
grayImage2.setRGB(row+bufferedImage1.getWidth()+blankWidth, col, bufferedImage2.getRGB(row, col));
}
}
charImages.set(i, grayImage2);
charImages.remove(i+1);
i++;
}
}
}
return charImages;
}
private double splitThreshold = 1.4;
/**
* 调小分割的间隙,再进行分割
* @param bufferedImage
* @return
*/
private List<BufferedImage> splitChar4(BufferedImage bufferedImage) {
List<BufferedImage> charImages = new ArrayList<>();
int width = bufferedImage.getWidth();
int height = bufferedImage.getHeight();
// 横坐标上的映射图片,如果x坐标上的像素个数小于这个阈值,就认为是噪点,舍弃掉,值越大,达到分割的间距越小,分的越细
//int threshold = 2;
// 字符间距,值越小,达到分割的间距越小,分的越细
//int interval = 1;
// 横坐标上的映射图片,如果x坐标上的像素个数小于这个阈值,就认为是噪点,舍弃掉,值越大,达到分割的间距越小,分的越细
int threshold = 2;
// 字符间距,值越小,达到分割的间距越小,分的越细
int interval = 1;
// 存储黑色像素在x轴上的映射,key为x轴坐标,value为对应x坐标上的黑色像素个数
Map<Integer, Integer> map = new HashMap<>();
// 存储字符起始和结束下标,key为起始下标,value为结束下标
Map<Integer, Integer> charsMap = new LinkedHashMap<>();
// 统计
for (int i = 0; i < width; i++) {
int blackCount = 0;
for (int j = 0; j < height; j++) {
// getRGB()方法,根据手册,其返回的int型数据(32位)为ARGB格式,其中ARGB各占8bit
int color = bufferedImage.getRGB(i, j);
int b = color & 0xff;
if (b == 0) {
blackCount++;
}
}
map.put(i, blackCount);
}
// 从左到右遍历时,记录白色像素点的宽度,超过这个值,就表示当前处于字符之间的空白处
int whiteCount = 0;
int startIndex = -1;
// 单个字符的最小宽度,用于判断当前字符是否需要合并
// 存在“出”字被分成两个部分,左边宽度大于bufferedImage.getHeight()*2/3,右边宽度为1像素,导致不能合并,需要增大minWidth
//int minWidth = bufferedImage.getHeight()*2/3;
int minWidth = bufferedImage.getHeight();
// 单个字符的最大宽度
// 存在“呀,”合并的情况,因为它们的宽度小于bufferedImage.getHeight()*5/4,需要缩小他的值
//int maxWidth = bufferedImage.getHeight()*5/4;
int maxWidth = height;
int lastKey = -1;
int lastValue = -1;
// 下标从左到右顺序:lastKey lastValue startIndex i
// 字符切割逻辑,从左往右依次扫描,碰到第一个黑色像素,标记为startIndex,继续扫描,碰到第一个白色像素,标记为i,
// 这时认为i-startIndex为第一个字符的宽度,如果此宽度大于设定的参数,就认为他是一个完整的字符,添加进去,重置startIndex变量
// 如果小于设定的参数,就认为他不是一个完整的字符,将startIndex赋值给lastKey,i赋值给lastValue,继续扫描,
// 当扫描到第二个字符的起始黑色像素点时,标记为startIndex,继续扫描,碰到第一个白色像素,标记为i,
// 如果第二个字符是完整的字符,即宽度超过设定值,那么第一个不完整的字符肯定是完整的字符,因为没有找到他的剩余部分
// 如果第二个字符也不是完整的字符,那么计算第二个字符的宽度加上之前第一个不完整字符的宽度即i-lastKey,
// 如果超过了设定的参数,就认为他们两个可以组成完整的字符,否则继续扫描剩余的部分
// 总结起来就是,我从左往右扫描,当扫描到的黑色像素块超过我设定的阈值,那么我就认为这些黑色像素块是一个字符,以此类推
// 找出起始像素点
for (int i = 0; i < width; i++) {
int value = map.get(i);
// 字符起始下标
if (startIndex == -1) {
// 对于高度小于设定阈值的坐标,直接舍弃
if (value >= threshold) {
startIndex = i;
}
} else {
// x轴上像素个数小于3的都认为是空白
if (value < threshold) {
whiteCount++;
} else {
// 对于一个字符中间存在空白的情况,需要排除掉
whiteCount = 0;
}
if (whiteCount > interval) {
int charWidth = i-interval-startIndex;
// 防止一个字符被分割成两个部分
if (charWidth < minWidth) {
// 如果是类似“儿”这样的左右分开的字符
if (lastKey == -1) {
lastKey = startIndex;
lastValue = i-interval;
} else {
if (i-interval-lastKey > maxWidth) {
charsMap.put(lastKey, lastValue);
lastKey = startIndex;
lastValue = i-interval;
} else {
/*charsMap.put(lastKey, i-interval);
lastKey = -1;
lastValue = -1;*/
// 存在一个字符被分割成多个情况,这时需要多次合并
lastValue = i-interval;
}
}
} else {
if (lastKey != -1) {
charsMap.put(lastKey, lastValue);
lastKey = -1;
lastValue = -1;
}
charsMap.put(startIndex, i-interval);
}
startIndex = -1;
whiteCount = 0;
}
}
}
// 处理行尾的字符
if (lastKey != -1 && startIndex != -1) {
int i = width;
// 后一个字符的宽度
int charWidth = i-interval-startIndex;
// 防止一个字符被分割成两个部分
if (charWidth < minWidth) {
// 如果是类似“儿”这样的左右分开的字符
if (i-interval-lastKey > maxWidth) {
charsMap.put(lastKey, lastValue);
charsMap.put(startIndex, i-interval);
} else {
// 存在一个字符被分割成多个情况,这时需要多次合并
charsMap.put(lastKey, i-interval);
}
} else {
charsMap.put(lastKey, lastValue);
charsMap.put(startIndex, i-interval);
}
lastKey = -1;
lastValue = -1;
startIndex = -1;
}
if (lastKey != -1 && startIndex == -1) {
// 过滤噪点,暂时不考虑与最后一个字符合并
if (lastValue - lastKey > 1) {
charsMap.put(lastKey, lastValue);
lastKey = -1;
lastValue = -1;
}
}
if (startIndex != -1 && lastKey == -1) {
charsMap.put(startIndex, width);
startIndex = -1;
}
// 根据上面已经标注好的字符下标,提取字符图片
for (Integer key : charsMap.keySet()) {
int value = charsMap.get(key);
if (value-key == 0) {
continue;
}
BufferedImage grayImage = bufferedImage.getSubimage(key,0,value-key,height);
/*BufferedImage grayImage = new BufferedImage(value-key, bufferedImage.getHeight(), bufferedImage.getType());
for (int i = 0; i < value-key; i++) {
for (int j = 0; j < bufferedImage.getHeight(); j++) {
int color = bufferedImage.getRGB(key+i, j);
grayImage.setRGB(i, j, color);
}
}*/
// 清除无效字符
if (isInvalidChar(grayImage)) {
continue;
}
// 清除字符上下空白部分
BufferedImage cutUDImage = cutUD(grayImage);
int maxHeight = getMaxHeight(bufferedImage);
if (cutUDImage.getWidth() > maxHeight*splitThreshold) {
// 如果两个字符相互重合不能使用一条直线分割,使用连通量进行分割
List<BufferedImage> list2 = splitChar5(cutUDImage);
charImages.addAll(list2);
} else {
charImages.add(cutUDImage);
}
}
return charImages;
}
// 用于处理连在一起的字符串
private int[] array2;
/**
* 根据连通分量来分割字符
* 只处理含有两个连通分量的图片
* @param image
* @return
*/
private List<BufferedImage> splitChar5(BufferedImage image) {
int width = image.getWidth();
int height = image.getHeight();
int size = width * height;
// 最小连通分量的面积,小于阈值的过滤掉
int threshold = 20;
// 用于存储每个单元格所属的集合
array2 = new int[size];
// 初始化各个单元格所属的集合
for (int j = 0; j < size; j++) {
array2[j] = -1;
}
//long startTime = System.currentTimeMillis();
// 先合并左右单元格中间的竖着的墙
for (int col=0; col<height; col++) {
for (int row=0; row<width-1; row++) {
int firstColor = image.getRGB(row, col);
int secondColor = image.getRGB(row+1, col);
int firstIndex = col*width+row;
if ((firstColor & 0xff) == 0 && firstColor == secondColor) {
// 将两个单元格连通
array2[firstIndex] = firstIndex+1;
}
}
}
// 再拆除上下单元格之间的横着的墙,只拆除上下两行最右边的两个单元格就行了,能保证两行可以合并到同一个联通分量里面就行
for (int col=height-1; col>0; col--) {
for (int row=width-1; row>-1; row--) {
int firstColor = image.getRGB(row, col);
int secondColor = image.getRGB(row, col-1);
// 上下两个单元格的右边如果都是黑色,就不合并
if ((firstColor & 0xff) == 0 && firstColor == secondColor) {
// 第一个像素在第二个像素的下面
int firstIndex = col*width+row;
int secondIndex = firstIndex - width;
union2(firstIndex, secondIndex);
}
}
}
// 统计所有连通分量的大小
Map<Integer, Integer> map = new HashMap<>();
// 将联通分量里面的元素全部设置为统一编号,便于统计,用倒序比正序快得多
for (int i=size-1; i>-1; i--) {
int row = array2[i];
if (row == -1) {
continue;
}
int result1 = find2(i);
Integer key = map.get(result1);
if (key == null) {
map.put(result1, 1);
} else if (key < threshold) {
map.put(result1, key+1);
}
array2[i] = result1;
}
// 存储连通量的上下两个极值在数组中的位置,key为连通量编号,也是连通量的上极值,value为连通量的下极值
Map<Integer, Integer> map2 = new LinkedHashMap<>();
for (int minY=0; minY<size; minY++) {
int maxY = array2[minY];
if (maxY == -1) {
continue;
}
// 过滤掉面积小于1500像素的联通分量
if (!map2.containsKey(maxY) && map.get(maxY) == threshold) {
map2.put(maxY, minY);
}
}
List<BufferedImage> list = new ArrayList<>();
if (map2.size() != 2) {
list.add(image);
return list;
}
// 根据联通分量来抠图
Map<Integer, BufferedImage> map5 = new LinkedHashMap<>();
for (int row = 0; row < width; row++) {
for (int col=0; col<height; col++) {
int index = col * width + row;
int maxY = array2[index];
if (maxY == -1 || map2.get(maxY) == null) {
continue;
}
int minY = map2.get(maxY);
if (map2.containsKey(maxY)) {
int newHeight = maxY/width - minY/width+1;
int blankHeight = minY/width;
if (map5.containsKey(maxY)) {
BufferedImage grayImage = map5.get(maxY);
int color = image.getRGB(row, col);
grayImage.setRGB(row, col-blankHeight, color);
} else {
// 找出能放下字符行的四边形的四个角
BufferedImage grayImage = new BufferedImage(width, newHeight, image.getType());
// 设置白色背景
for (int i=0; i<width; i++) {
for (int j=0; j<newHeight; j++) {
grayImage.setRGB(i, j, 0xFFFFFF);
}
}
int color = image.getRGB(row, col);
grayImage.setRGB(row, col-blankHeight, color);
map5.put(maxY, grayImage);
}
}
}
}
for (Integer key : map5.keySet()) {
// 清除字符上下空白部分
BufferedImage cutLR2 = cutLR2(map5.get(key));
list.add(cutLR2);
}
return list;
}
private boolean isInvalidChar(BufferedImage bufferedImage) {
int blackCount = 0;
// 统计
for (int i = 0; i < bufferedImage.getWidth(); i++) {
for (int j = 0; j < bufferedImage.getHeight(); j++) {
// getRGB()方法,根据手册,其返回的int型数据(32位)为ARGB格式,其中ARGB各占8bit
int color = bufferedImage.getRGB(i, j);
int b = color & 0xff;
if (b == 0) {
blackCount++;
}
}
}
return blackCount < ignoreArea;
}
/**
* 获取一行文字的最大有效高度
* 用于处理倒S形的文字行
* 从左到右扫描列,通过记录每一列的第一个和最后一个黑色像素,从而得到该列的有效高度,最后取所有列最大的有效高度,也即是字符的最大高度
* @param bufferedImage
* @return
*/
private static int getMaxHeight(BufferedImage bufferedImage) {
int maxHeight = 0;
// 以图片左上角点为坐标原点
for (int i = 0; i < bufferedImage.getWidth(); i++) {
int start = bufferedImage.getHeight()-1;
int end = 0;
for (int j = 0; j < bufferedImage.getHeight(); j++) {
int color = bufferedImage.getRGB(i, j);
int red = color & 0xFF;
if (red == 0) {
start = j;
break;
}
}
for (int j = bufferedImage.getHeight()-1; j > -1; j--) {
int color = bufferedImage.getRGB(i, j);
int red = color & 0xFF;
if (red == 0) {
end = j;
break;
}
}
if ((end - start) > maxHeight) {
maxHeight = end - start;
}
}
return maxHeight + 1;
}
/**
* 分割图片
* 在原图上进行分割,不重新创建BufferedImage对象
* @param bufferedImage
* @return
*/
private static Map<Integer, BufferedImage> splitImage2(BufferedImage bufferedImage) {
Map<Integer, BufferedImage> map = new HashMap<>();
int w = bufferedImage.getWidth();
int h = bufferedImage.getHeight();
int blockWidth = w / widthSplitNum;
int blockHeight = h / heightSplitNum;
for (int i = 0; i < widthSplitNum; i++) {
for (int j = 0; j < heightSplitNum; j++) {
int mStart = i * blockWidth;
int nStart = j * blockHeight;
BufferedImage grayImage2 = bufferedImage.getSubimage(mStart, nStart, blockWidth, blockHeight);
map.put(i * heightSplitNum + j, grayImage2);
}
}
return map;
}
public BufferedImage mergeImage(Map<Integer, BufferedImage> map) {
if (map == null || map.get(0) == null) {
return null;
}
int blockWidth = map.get(0).getWidth();
int blockHeight = map.get(0).getHeight();
BufferedImage destImage = new BufferedImage(blockWidth * widthSplitNum, blockHeight * heightSplitNum, BufferedImage.TYPE_BYTE_BINARY);
for (int i = 0; i < widthSplitNum; i++) {
for (int j = 0; j < heightSplitNum; j++) {
BufferedImage grayImage = map.get(i * heightSplitNum + j);
for (int m = 0; m < grayImage.getWidth(); m++) {
for (int n = 0; n < grayImage.getHeight(); n++) {
destImage.setRGB(i * blockWidth + m, j * blockHeight + n, grayImage.getRGB(m, n));
}
}
}
}
return destImage;
}
/**
* 将单文字图片组合成指定宽度的文字图片
* 每一行文字, 按中间对齐, 均匀分布在一行中
* @param list 单文字图片列表
* @param width 组合图片宽度
* @param colSpace 列间距
* @param rowSpace 行间距
* @return
*/
public BufferedImage mergeImage5(List<BufferedImage> list, int width, int colSpace, int rowSpace) {
if (list == null || list.isEmpty()) {
return null;
}
// 当前行单文字组合的宽度
int currentWidth = 0;
// 存放每一行的单文字图片
Map<Integer, List<BufferedImage>> picMap = new HashMap<>();
// 记录每一行文字距离顶部的高度
Map<Integer, Integer> rowsHeight = new HashMap<>();
rowsHeight.put(0, rowSpace);
// 行下标
int rowIndex = 0;
// 单行文字最高高度
int maxHeight = 0;
// 组合文字图片高度
int totalHeight = rowSpace;
// 计算文字与文字之间的间距, 对应text-align: justify; 向两侧对齐,最后一行不适用
Map<Integer, Integer> charBlankMap = new HashMap<>();
// 每行文字图片的宽度和
int picsWidth = 0;
List<BufferedImage> picList = new ArrayList<>();
for (BufferedImage image : list) {
// 首行缩进换行
BufferedImage temp = cutL(image);
if (image.getWidth() - textIndent == temp.getWidth()) {
charBlankMap.put(rowIndex, colSpace);
picsWidth = 0;
picMap.put(rowIndex++, picList);
totalHeight += rowSpace + maxHeight;
rowsHeight.put(rowIndex, totalHeight);
currentWidth = 0;
maxHeight = 0;
picList = new ArrayList<>();
currentWidth += image.getWidth() + colSpace;
maxHeight = image.getHeight();
picList.add(image);
picsWidth += image.getWidth();
continue;
}
int charWidth = image.getWidth() + colSpace;
// 超出行长度, 自动换行
if (currentWidth + charWidth > width) {
charBlankMap.put(rowIndex, (width-picsWidth)/(picList.size()-1));
//charBlankMap.put(rowIndex, colSpace);
picsWidth = 0;
picMap.put(rowIndex++, picList);
totalHeight += rowSpace + maxHeight;
rowsHeight.put(rowIndex, totalHeight);
currentWidth = 0;
maxHeight = 0;
picList = new ArrayList<>();
}
picsWidth += image.getWidth();
currentWidth += charWidth;
picList.add(image);
if (image.getHeight() > maxHeight) {
maxHeight = image.getHeight();
}
// 行末尾空白换行
BufferedImage temp2 = cutR(image);
if (image.getWidth() - rowBlankWidth == temp2.getWidth()) {
charBlankMap.put(rowIndex, colSpace);
picsWidth = 0;
picMap.put(rowIndex++, picList);
totalHeight += rowSpace + maxHeight;
rowsHeight.put(rowIndex, totalHeight);
currentWidth = 0;
maxHeight = 0;
picList = new ArrayList<>();
}
}
// 最后一行
if (currentWidth > 0) {
charBlankMap.put(rowIndex, colSpace);
picsWidth = 0;
picMap.put(rowIndex++, picList);
totalHeight += rowSpace + maxHeight;
rowsHeight.put(rowIndex, totalHeight);
}
// colSpace*10 图片左右两边空白边界
BufferedImage destImage = new BufferedImage(width + colSpace*10, totalHeight + rowSpace*2, BufferedImage.TYPE_BYTE_BINARY);
// 设置白色背景
for (int i=0; i<destImage.getWidth(); i++) {
for (int j=0; j<destImage.getHeight(); j++) {
destImage.setRGB(i, j, 0xFFFFFF);
}
}
for (Integer index: picMap.keySet()) {
// 拼接行时, 记录行宽度下标
int tempWidthIndex = 0;
for (BufferedImage grayImage : picMap.get(index)) {
// 对于"一"字,需要放在中间
int tempHeight = 0;
//if (grayImage.getHeight() < 11) {
tempHeight = (rowsHeight.get(index+1)-rowsHeight.get(index) - rowSpace)/2 - grayImage.getHeight()/2;
//}
for (int m = 0; m < grayImage.getWidth(); m++) {
for (int n = 0; n < grayImage.getHeight(); n++) {
int rgb = grayImage.getRGB(m, n);
// colSpace*5 图片左边空白界
try {
destImage.setRGB(tempWidthIndex + m + colSpace * 5, rowsHeight.get(index) + n + tempHeight, rgb);
} catch (Exception e) {
e.printStackTrace();
}
}
}
tempWidthIndex += charBlankMap.get(index) + grayImage.getWidth();
}
}
return destImage;
}
/**
* 二值化图片
*
* @param bufferedImage 原图片
* @return 二值化后的图片
*/
private static BufferedImage binaryImage(BufferedImage bufferedImage) {
BufferedImage grayImage = new BufferedImage(bufferedImage.getWidth(), bufferedImage.getHeight(), bufferedImage.getType());
int threshold = getMeanThreshold(bufferedImage);
for (int i = 0; i < bufferedImage.getWidth(); i++) {
for (int j = 0; j < bufferedImage.getHeight(); j++) {
// getRGB()方法,根据手册,其返回的int型数据(32位)为ARGB格式,其中ARGB各占8bit
int color = bufferedImage.getRGB(i, j);
int r = (color >> 16) & 0xff;
int g = (color >> 8) & 0xff;
int b = color & 0xff;
int gray = (int) (0.2126 * r + 0.7152 * g + 0.0722 * b);
if (gray > threshold) {
// 白色
grayImage.setRGB(i, j, 0xFFFFFF);
} else {
// 黑色
grayImage.setRGB(i, j, 0);
}
}
}
return grayImage;
}
/**
* 获取图片的阀值,采用基于灰度平均值的阈值
*
* @param bufferedImage 原图片
* @return 二值化的阈值
*/
private static int getMeanThreshold(BufferedImage bufferedImage) {
double aa = 0.83;
int w = bufferedImage.getWidth();
int h = bufferedImage.getHeight();
int num = 0;
long sum = 0;
for (int i = 0; i < w; i++) {
for (int j = 0; j < h; j++) {
int color = bufferedImage.getRGB(i, j);
int r = (color >> 16) & 0xff;
int g = (color >> 8) & 0xff;
int b = color & 0xff;
int gray = (int) (0.2126 * r + 0.7152 * g + 0.0722 * b);
sum += gray;
num += 1;
}
}
// 测试表明,阀值取平均值的1.2倍效果最好。
// 越大越黑
int threshold = (int) (sum / num);
if (threshold * aa < 255) {
threshold = (int) (aa * sum / num);
}
return threshold;
}
// 用于存储每个单元格所属的集合
private int[] array;
// 映射到右边边界上,用于判断哪些行有黑色像素,对于没有黑色像素的行直接跳过
private Map<Integer, Integer> map3;
/**
* 减少union的时间
* @param image
* @return
* @throws IOException
*/
private List<BufferedImage> getCharRow4(BufferedImage image, BufferedImage bufferedImage) {
int width = image.getWidth();
int height = image.getHeight();
int size = width * height;
// 最小连通分量的面积,小于阈值的过滤掉
int threshold = 800;
// 用于存储每个单元格所属的集合
array = new int[size];
// 初始化各个单元格所属的集合
for (int j = 0; j < size; j++) {
array[j] = -1;
}
//long startTime = System.currentTimeMillis();
// 先合并左右单元格中间的竖着的墙
for (int col=0; col<height; col++) {
if (!map3.containsKey(col)) {
continue;
}
for (int row=0; row<width-1; row++) {
int firstColor = image.getRGB(row, col);
int secondColor = image.getRGB(row+1, col);
int firstIndex = col*width+row;
if ((firstColor & 0xff) == 0 && firstColor == secondColor) {
// 将两个单元格连通
array[firstIndex] = firstIndex+1;
}
}
}
boolean flag = false;
// 再拆除上下单元格之间的横着的墙,只拆除上下两行最右边的两个单元格就行了,能保证两行可以合并到同一个联通分量里面就行
for (int col=height-1; col>0; col--) {
if (!map3.containsKey(col)) {
continue;
}
// 每次换行都要执行一次合并,防止因为两行的最右边是边界,没有白色像素
flag = true;
for (int row=width-1; row>-1; row--) {
int firstColor = image.getRGB(row, col);
int secondColor = image.getRGB(row, col-1);
// 如果两个黑色像素前面是白色像素,那么就表示他们是各自行的末尾,那么就进行合并操作
if (!flag && (firstColor == -1 || secondColor == -1)) {
flag = true;
}
// 上下两个单元格的右边如果都是黑色,就不合并
if (flag && ((firstColor & 0xff) == 0 && firstColor == secondColor)) {
// 第一个像素在第二个像素的下面
int firstIndex = col*width+row;
int secondIndex = firstIndex - width;
union(firstIndex, secondIndex);
flag = false;
}
}
}
//System.out.println("startTime3:" + (System.currentTimeMillis()- startTime) + "ms");
// 统计所有连通分量的大小
Map<Integer, Integer> map = new HashMap<>();
// 将联通分量里面的元素全部设置为统一编号,便于统计,用倒序比正序快得多
for (int i=size-1; i>-1; i--) {
int row = array[i];
if (row == -1) {
continue;
}
int result1 = find(i);
Integer key = map.get(result1);
if (key == null) {
map.put(result1, 1);
} else if (key < threshold) {
map.put(result1, key+1);
}
array[i] = result1;
}
// 存储连通量的上下两个极值在数组中的位置,key为连通量编号,也是连通量的上极值,value为连通量的下极值
Map<Integer, Integer> map2 = new LinkedHashMap<>();
for (int minY=0; minY<size; minY++) {
int maxY = array[minY];
if (maxY == -1) {
continue;
}
// 过滤掉面积小于1500像素的联通分量
if (!map2.containsKey(maxY) && map.get(maxY) == threshold) {
map2.put(maxY, minY);
}
}
//System.out.println("startTime6:" + (System.currentTimeMillis()- startTime) + "ms");
// 根据联通分量来抠图
List<BufferedImage> list = new ArrayList<>();
Map<Integer, BufferedImage> map5 = new LinkedHashMap<>();
for (int col=0; col<height; col++) {
if (!map3.containsKey(col)) {
continue;
}
for (int row = 0; row < width; row++) {
int index = col * width + row;
int maxY = array[index];
if (maxY == -1 || map2.get(maxY) == null) {
continue;
}
int minY = map2.get(maxY);
if (map2.containsKey(maxY)) {
int newHeight = maxY/width - minY/width+1;
int blankHeight = minY/width;
if (map5.containsKey(maxY)) {
BufferedImage grayImage = map5.get(maxY);
int color = bufferedImage.getRGB(row, col);
grayImage.setRGB(row, col-blankHeight, color);
} else {
// 找出能放下字符行的四边形的四个角
BufferedImage grayImage = new BufferedImage(width, newHeight, image.getType());
// 设置白色背景
for (int i=0; i<width; i++) {
for (int j=0; j<newHeight; j++) {
grayImage.setRGB(i, j, 0xFFFFFF);
}
}
int color = bufferedImage.getRGB(row, col);
grayImage.setRGB(row, col-blankHeight, color);
map5.put(maxY, grayImage);
}
}
}
}
for (Integer key : map5.keySet()) {
list.add(map5.get(key));
}
return list;
}
public String fillBlank(String str, int length) {
int strLength = str.length();
if (strLength < length) {
for (int i=0; i<length-strLength; i++) {
str += " ";
}
}
return str;
}
/**
* 返回 i 所在集合的最大值
*
* @param i 单元格编号
* @return
*/
public int find(int i) {
int result = i;
while (array[result] != -1) {
result = array[result];
}
return result;
}
/**
* 将 i 和 j 所在集合进行合并
*
* @param i 单元格编号
* @param j 单元格编号
*/
public void union(int i, int j) {
int result1 = find(i);
int result2 = find(j);
if (result1 == result2){
return;
}
if(result1 > result2) {
array[result2] = result1;
}
else {
array[result1] = result2;
}
}
public int find2(int i) {
int result = i;
while (array2[result] != -1) {
result = array2[result];
}
return result;
}
public void union2(int i, int j) {
int result1 = find2(i);
int result2 = find2(j);
if (result1 == result2){
return;
}
if(result1 > result2) {
array2[result2] = result1;
}
else {
array2[result1] = result2;
}
}
}