Tesseract-OCR Windows下Java代码实现

一、项目背景

        基于图片文字识别的需求，在系统中实现此功能。

二、实现思路

        前台上传需要识别的图片，后台接收到图片进行识别，将识别结果（文字字符串）返回给前台，本文着重介绍识别部分的简单实现，上传功能不做介绍。

三、实现过程

        新问题往往需要使用新的技术去解决，过程也许是曲折艰辛的，但结果总是证明了我们的成长。直接进入正题吧，本文使用Tesseract-OCR，官方地址https://code.google.com/p/tesseract-ocr/，Maven仓库可参考http://mvnrepository.com/artifact/net.sourceforge.tess4j/tess4j/3.2.0

1.新建maven项目，在pom.xml中添加依赖包，并在lib路径下添加.dll文件

2.将.dll文件添加到windows系统中,将jai-imageio-core.jar加到jdk的lib中

3.项目中新建tessdata文件夹，放置字库等信息

（eng.traineddata文件对应英文字库，可将下载好的中文字库chi_sim.traineddata重命名替换eng.traineddata,就可解析中文，当然也有其他办法，可自行研究）

以上相当于配置tesseract-ocr环境

4.识别代码

   /**
	 * 提交图片解析
	 * @param bpmProspectingStr 
     *                          其他一并提交参数
	 * @return
	 */
	@ResponseBody
	@RequestMapping(value = "mSave")
	public Map<String,Object> mSave(HttpServletRequest request, HttpServletResponse response,String bpmProspectingStr) throws IOException {
		Map<String, Object> map = Maps.newHashMap();
		List<SystemInfo> result = Lists.newArrayList();
		MultipartFile multipartFile = null;
		CommonsMultipartResolver multipartResolver = new CommonsMultipartResolver(request.getSession().getServletContext());
		if(!multipartResolver.isMultipart(request)) { //判断request是否有文件上传
			result.add(new SystemInfo("error", AppInfo.getStateInfo(AppInfo.NOT_MULTIPART_CONTENT)));
			return map;
		}
		//上传图片
		String photoFileName =  UploadUtil.mobileUpload(request, response, configJSONPath);

		String contentPath = request.getContextPath();
		String realPath = request.getSession().getServletContext().getRealPath(contentPath);
		realPath = realPath.substring(0,realPath.lastIndexOf("smartinspection"));
		String path="static/upload/";
		realPath = realPath+path+photoFileName;
		MultipartHttpServletRequest multiRequest = (MultipartHttpServletRequest) request;  //multipartResolver.resolveMultipart(request);
		Iterator<String> iterator = multiRequest.getFileNames();
		while (iterator.hasNext()) {
			multipartFile = multiRequest.getFile(iterator.next()); 
			if (multipartFile != null) {
				result.add(new SystemInfo("error", AppInfo.getStateInfo(AppInfo.NOTFOUND_UPLOAD_DATA)));
				String reStr = getLocation(realPath);
				map.put("reStr", reStr);
				break;
			}else{
				continue;
			}
		}
		map.put("status", "true");
		return map;
	}

	private String getLocation(String filePath) throws IOException {
		BufferedImage image = null;
		File imagefile = null;
		//you can either use URL or File for reading image using ImageIO
		imagefile = new File(filePath);
		File dataFile = imagefile;
		File tempImage = ImageIOHelper.createImage(dataFile, "jpg");
		if(dataFile.exists()){
			ImageIO.scanForPlugins();
			Tesseract instance = Tesseract.getInstance(); // JNA Interface Mapping
			System.err.println("检验环境是否支持TIFF");
			String format="tif";
			Iterator<ImageReader> readers = ImageIO.getImageReadersByFormatName(format);
			System.out.println(readers.hasNext());
			Iterator<ImageWriter> writers = ImageIO.getImageWritersByFormatName("tiff");
			System.out.println(writers.hasNext());
			try {
				capctha = instance.doOCR(tempImage).trim();
				System.err.println("解析结果capctha is： "+capctha);
			} catch (TesseractException e) {
				e.printStackTrace();
			}
		} else {
			System.err.println("找不到文件");
		}
		return capctha;
	}

 

转载于:https://my.oschina.net/u/589182/blog/703246