@PostConstruct
public void initOcrEngine
() {
tesseract = new Tesseract
();
//语言包路径和支持语言
tesseract
.setDatapath
(tessDataPath
);
tesseract
.setLanguage
("eng+chi_sim"
);
tesseract
.setPageSegMode
(6
); //自动页面分割
tesseract
.setOcrEngineMode
(3
); //LSTM引擎
}
/**
* OCR识别图片内容
*/
private String extractImageText
(MultipartFile file
) {
try
(InputStream is = file
.getInputStream
()) {
BufferedImage image = ImageIO
.read
(is
);
if
(image == null
) {
return MessageUtils
.message
("Image
.parsing
.failed"
);
}
image = binarizeImage
(image, 128
); // 二
值化
image = breakCharacterConnections
(image
); // 断开粘连
List<Rectangle> regions = findCharacterRegions
(image
); // 字符级检测
image = addSpacingBetweenCharacters
(image, regions, 10
); // 插入间距
saveDebugImage
(image, "3_final
.png"
); // 保存最终处理结果
String result = tesseract
.doOCR
(image
); //OCR识别
result = postProcess
(result
);
result = result
.replaceAll
("\\s+", " "
).trim
();
System
.out
.pr
intln
("图片内容：\n" + result
);
return result;
} catch
(Exception e
) {
e
.pr
intStackTrace
();
return MessageUtils
.message
("file
.read
.picture
.error"
);
}
}
private void saveDebugImage
(BufferedImage img, String filename
) throws IOException {
// 保存到项目目录下的debug文件夹
String basePath = "D:/pdf/debug/";
File outputDir = new File
(basePath
);
if
(!outputDir
.exists
()) outputDir
.mkdirs
();
String fullPath = basePath + filename;
ImageIO
.write
(img, "png", new File
(fullPath
));
}
private static BufferedImage binarizeImage
(BufferedImage image,
int threshold
) {
int width = image
.getWidth
();
int height = image
.getHeight
();
BufferedImage binaryImage = new BufferedImage
(width, height, BufferedImage
.TYPE_BYTE_BINARY
);
for
(int y = 0; y < height; y++
) {
for
(int x = 0; x < width; x++
) {
int a
rgb = image
.get
RGB(x, y
);
int gray =
(a
rgb >> 16
) & 0xff;
if
(gray < threshold
) {
binaryImage
.setRGB(x, y,
Color.BLACK
.get
RGB());
} else {
binaryImage
.setRGB(x, y,
Color.WHITE
.get
RGB());
}
}
}
return binaryImage;
}
public static List<Rectangle> findCharacterRegions
(BufferedImage binaryImage
) {
int width = binaryImage
.getWidth
();
int height = binaryImage
.getHeight
();
boolean[][] visited = new boolean[width][height];
List<Rectangle> regions = new ArrayList<>
();
for
(int y = 0; y < height; y++
) {
for
(int x = 0; x < width; x++
) {
if
(isBlack
(binaryImage
.get
RGB(x, y
)) && !visited[x][y]
) {
// BFS查找连通域
Queue<Po
int> queue = new LinkedList<>
();
queue
.add
(new Po
int(x, y
));
visited[x][y] = true;
int minX = x, maxX = x, minY = y, maxY = y;
while
(!queue
.isEmpty
()) {
Po
int p = queue
.poll
();
for
(int dx =
-1; dx <= 1; dx++
) {
for
(int dy =
-1; dy <= 1; dy++
) {
int nx = p
.x + dx, ny = p
.y + dy;
if
(nx >= 0 && nx < width && ny >= 0 && ny < height
&& isBlack
(binaryImage
.get
RGB(nx, ny
)) && !visited[nx][ny]
) {
visited[nx][ny] = true;
queue
.add
(new Po
int(nx, ny
));
minX = Math
.min
(minX, nx
);
maxX = Math
.max
(maxX, nx
);
minY = Math
.min
(minY, ny
);
maxY = Math
.max
(maxY, ny
);
}
}
}
}
// 过滤小面积噪点
if
((maxX
- minX + 1
) *
(maxY
- minY + 1
) > 20
) {
regions
.add
(new Rectangle
(minX, minY, maxX
- minX + 1, maxY
- minY + 1
));
}
}
}
}
return regions;
}
private static boolean isBlack
(int rgb) {
return
(rgb & 0xFFFFFF
) == 0; // 纯黑色
}
public static BufferedImage addSpacingBetweenCharacters
(BufferedImage image, List<Rectangle> regions,
int spacing
) {
int width = image
.getWidth
();
int height = image
.getHeight
();
BufferedImage newImage = new BufferedImage
(width + spacing * regions
.size
(), height, BufferedImage
.TYPE_BYTE_BINARY
);
Graphics2D g = newImage
.create
Graphics();
g
.setColor(Color.WHITE
);
g
.fillRect
(0, 0, newImage
.getWidth
(), newImage
.getHeight
());
int offsetX = 0;
for
(Rectangle r : regions
) {
// 截取字符区域
BufferedImage charImg = image
.getSubimage
(r
.x, r
.y, r
.width, r
.height
);
g
.drawImage
(charImg, offsetX, 0, null
);
offsetX += r
.width + spacing; // 插入间距
}
g
.dispose
();
return newImage;
}
public static BufferedImage breakCharacterConnections
(BufferedImage image
) {
int width = image
.getWidth
();
int height = image
.getHeight
();
BufferedImage result = new BufferedImage
(width, height, BufferedImage
.TYPE_BYTE_BINARY
);
// 定义3x3结构元素
int[][] kernel = {
{1, 1, 1},
{1, 1, 1},
{1, 1, 1}
};
// 腐蚀（缩小边缘）
BufferedImage eroded = new BufferedImage
(width, height, BufferedImage
.TYPE_BYTE_BINARY
);
for
(int y = 1; y < height
- 1; y++
) {
for
(int x = 1; x < width
- 1; x++
) {
boolean allBlack = true;
for
(int dy =
-1; dy <= 1; dy++
) {
for
(int dx =
-1; dx <= 1; dx++
) {
if
((image
.get
RGB(x + dx, y + dy
) & 0xFF
) > 0x7F
) {
allBlack = false;
break;
}
}
}
eroded
.setRGB(x, y, allBlack
? Color.BLACK
.get
RGB() :
Color.WHITE
.get
RGB());
}
}
// 膨胀（恢复字符主体）
for
(int y = 1; y < height
- 1; y++
) {
for
(int x = 1; x < width
- 1; x++
) {
boolean anyBlack = false;
for
(int dy =
-1; dy <= 1; dy++
) {
for
(int dx =
-1; dx <= 1; dx++
) {
if
((eroded
.get
RGB(x + dx, y + dy
) & 0xFF
) == 0
) {
anyBlack = true;
break;
}
}
}
result
.setRGB(x, y, anyBlack
? Color.BLACK
.get
RGB() :
Color.WHITE
.get
RGB());
}
}
return result;
}对嘛
此博客展示了一段Java代码，利用Java的AWT和ImageIO库进行图像绘制。代码中创建了一个BufferedImage对象，使用Graphics的setColor方法设置颜色，绘制矩形和椭圆形，最后将图像转换为灰度图并保存为PNG文件。
扫一扫
616
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
