使用tess4j完成身份证和营业执照图片的文字识别

这两天研究了一下关于OCR图文解析的技术。当然市场上已经有开源服务，比如百度的AI开放平台，就有OCR相关的API接口。我这里选用的是Tesseract开源框架，java封装版本是tess4j。结合网上公布的一些开源项目提供的demo，完成了身份证与营业执照的相关文字识别的处理。总体上来讲Tesseract其实还不错，简单应用其实还挺简单的（提供的图片质量可以靠前端做好限制，比如身份证识别，加上头像或国徽的框图限定，能提高识别率）。

　　示例项目地址：https://github.com/git-simm/simm-framework

一、技术介绍

OCR(Optical Character Recognition)：光学字符识别，是指电子设备（例如扫描仪或数码相机）检查纸上打印的字符，通过检测暗、亮的模式确定其形状，然后用字符识别方法将形状翻译成计算机文字的过程。

Tesseract：开源的OCR识别引擎，初期Tesseract引擎由HP实验室研发，后来贡献给了开源软件业，后由Google进行改进、修改bug、优化，重新发布。

　　Tess4J：是对Tesseract OCR API.的Java JNA 封装。使java能够通过调用Tess4J的API来使用Tesseract OCR。支持的格式：TIFF,JPEG,GIF,PNG,BMP,JPEG,and PDF

Tess4J API 提供的功能：
1、直接识别支持的文件
2、识别图片流
3、识别图片的某块区域
4、将识别结果保存为 TEXT/ HOCR/ PDF/ UNLV/ BOX
5、通过设置取词的等级，提取识别出来的文字
6、获得每一个识别区域的具体坐标范围
7、调整倾斜的图片
8、裁剪图片
9、调整图片分辨率
10、从粘贴板获得图像
11、克隆一个图像（目的：创建一份一模一样的图片，与原图在操作修改上，不相 互影响）
12、图片转换为二进制、黑白图像、灰度图像
13、反转图片颜色

二、环境准备(https://www.jianshu.com/p/ef60ef5395c5)

　　2.1、我们需要安装tessdata语言包，用于图文识别。 tesseract-ocr语言包的下载地址，用于识别文字时进行匹配。链接: https://pan.baidu.com/s/1XAvPkTdUXuFq-q2InDREhQ 提取码: 6vjp

　　　 

　　2.2、项目引入maven依赖

<dependency>

            <groupId>net.sourceforge.tess4j</groupId>

            <artifactId>tess4j</artifactId>

            <version>4.5.2</version>

        </dependency>

三、简单描述下图文解析的过程

　　

四、服务端关键代码的展示

private static int targetBrightness = 260;

    private static int targetDifferenceValue = 15;

    /**

     * 解析身份证信息

     *

     * @param inputStream

     * @return

     * @throws Exception

     */

    @Override

    public BizLicenseInfo getInfo(InputStream inputStream) throws Exception {

        BizLicenseInfo bizLicenseInfo = new BizLicenseInfo();

        String rootPath = ClassUtils.getDefaultClassLoader().getResource("").getPath()+"/tmp";

        Tesseract tesseract = new Tesseract();

        tesseract.setLanguage("chi_sim");

        //读取网络图片

        BufferedImage bufferedImage = ImageFilter.cloneImage(ImageIO.read(inputStream));

        //不过滤部分颜色

        //bufferedImage = ImageFilter.imageRGBDifferenceFilter(bufferedImage, targetDifferenceValue, null);

        bufferedImage = ImageFilter.convertImageToGrayScale(bufferedImage);

        //缩放到真实身份证大小

        bufferedImage = ImageFilter.imageScale(bufferedImage, 3150, 1920);

        try (OutputStream outputStream = new FileOutputStream(rootPath+"/bg.jpg")) {

            saveImg(bufferedImage,outputStream);

            getBufferedNameImage(tesseract, bufferedImage, bizLicenseInfo,rootPath+"/nameImageBefore.jpg");

            getBufferedCapitalImage(tesseract, bufferedImage, bizLicenseInfo,rootPath+"/capitalImageBefore.jpg");

            getBufferedBizTypeImage(tesseract, bufferedImage, bizLicenseInfo,rootPath+"/bizTypeImageBefore.jpg");

            getBufferedBuildOnImage(tesseract, bufferedImage, bizLicenseInfo,rootPath+"/buildOnImageBefore.jpg");

            getBufferedJuridicalImage(tesseract, bufferedImage, bizLicenseInfo,rootPath+"/juridicalImageBefore.jpg");

            getBufferedBizLimitImage(tesseract, bufferedImage, bizLicenseInfo,rootPath+"/bizLimitImageBefore.jpg");

            getBufferedBizScopeImage(tesseract, bufferedImage, bizLicenseInfo,rootPath+"/bizScopeImageBefore.jpg");

            getBufferedAddressImage(tesseract, bufferedImage, bizLicenseInfo,rootPath+"/addressImageBefore.jpg");

            getBufferedCreditCodeImage(tesseract, bufferedImage, bizLicenseInfo,rootPath+"/creditCodeImageBefore.jpg");

            return bizLicenseInfo;

        }catch (Exception e){

            e.printStackTrace();

            throw e;

        }

    }

/**

     * 获取统一社会信用代码

     * @param tesseract

     * @param bufferedImage

     * @param bizLicenseInfo

     * @param path

     * @throws IOException

     * @throws TesseractException

     */

    private void getBufferedCreditCodeImage(Tesseract tesseract, BufferedImage bufferedImage, BizLicenseInfo bizLicenseInfo, String path) throws IOException, TesseractException {

        try (OutputStream outputStream = new FileOutputStream(path)) {

            BufferedImage idImage = ImageFilter.subImage(bufferedImage, bufferedImage.getMinX() + 200

                    , 250, 550, 300);

            System.out.println("creditCodeImage 辉度处理");

            handBrightness(idImage, targetBrightness);

            saveImg(idImage, outputStream);

//            tesseract.setLanguage("eng");

            tesseract.setLanguage("chi_sim");

            // \W 可以配置 非字母和数字，等价于 [^a-zA-Z0-9] (\d \D 小写表示匹配数字，大写表示匹配非数字)

            String idCardNumber = tesseract.doOCR(idImage).replaceAll("[\\W]", "");

            bizLicenseInfo.setCreditCode(idCardNumber);

        }catch (Exception e){

            e.printStackTrace();

            throw e;

        }

    }

    /**

     * 获取名称

     * @param tesseract

     * @param bufferedImage

     * @param bizLicenseInfo

     * @param path

     */

    private void getBufferedNameImage(Tesseract tesseract, BufferedImage bufferedImage, BizLicenseInfo bizLicenseInfo, String path) throws IOException, TesseractException {

        BufferedImage buffered = ImageFilter.subImage(bufferedImage, 520, 700, 1200, 120);

        getBufferedImage(tesseract,buffered,path,(img,content)->{

            System.out.println("setName 辉度处理");

            bizLicenseInfo.setName(content);

        });

    }

    /**

     * 获取类型

     * @param tesseract

     * @param bufferedImage

     * @param bizLicenseInfo

     * @param path

     * @throws IOException

     * @throws TesseractException

     */

    private void getBufferedBizTypeImage(Tesseract tesseract, BufferedImage bufferedImage, BizLicenseInfo bizLicenseInfo, String path) throws IOException, TesseractException {

        BufferedImage buffered = ImageFilter.subImage(bufferedImage, 520, 820, 1200, 130);

        getBufferedImage(tesseract,buffered,path,(img,content)->{

            System.out.println("setBizType 辉度处理");

            bizLicenseInfo.setBizType(content);

        });

    }

    /**

     * 获取法人信息

     * @param tesseract

     * @param bufferedImage

     * @param bizLicenseInfo

     * @param path

     * @throws IOException

     * @throws TesseractException

     */

    private void getBufferedJuridicalImage(Tesseract tesseract, BufferedImage bufferedImage, BizLicenseInfo bizLicenseInfo, String path) throws IOException, TesseractException {

        BufferedImage buffered = ImageFilter.subImage(bufferedImage, 520, 950, 1200, 120);

        getBufferedImage(tesseract,buffered,path,(img,content)->{

            System.out.println("setJuridical 辉度处理");

            bizLicenseInfo.setJuridical(content);

        });

    }

    /**

     * 获取经营范围

     * @param tesseract

     * @param bufferedImage

     * @param bizLicenseInfo

     * @param path

     * @throws IOException

     * @throws TesseractException

     */

    private void getBufferedBizScopeImage(Tesseract tesseract, BufferedImage bufferedImage, BizLicenseInfo bizLicenseInfo, String path) throws IOException, TesseractException {

        BufferedImage buffered = ImageFilter.subImage(bufferedImage, 520, 1070, 1330, bufferedImage.getHeight() - 1200);

        getBufferedImage(tesseract,buffered,path,(img,content)->{

            System.out.println("setBizScope 辉度处理");

            bizLicenseInfo.setBizScope(content);

        });

    }

    /**

     * 获取注册资本

     * @param tesseract

     * @param bufferedImage

     * @param bizLicenseInfo

     * @param path

     * @throws IOException

     * @throws TesseractException

     */

    private void getBufferedCapitalImage(Tesseract tesseract, BufferedImage bufferedImage, BizLicenseInfo bizLicenseInfo, String path) throws IOException, TesseractException {

        BufferedImage buffered = ImageFilter.subImage(bufferedImage, 2170, 720, bufferedImage.getWidth()-2400, 120);

        getBufferedImage(tesseract,buffered,path,(img,content)->{

            System.out.println("setCapital 辉度处理");

            bizLicenseInfo.setCapital(content);

        });

    }

    /**

     * 获取成立日期

     * @param tesseract

     * @param bufferedImage

     * @param bizLicenseInfo

     * @param path

     * @throws IOException

     * @throws TesseractException

     */

    private void getBufferedBuildOnImage(Tesseract tesseract, BufferedImage bufferedImage, BizLicenseInfo bizLicenseInfo, String path) throws IOException, TesseractException {

        BufferedImage buffered = ImageFilter.subImage(bufferedImage, 2170, 850, bufferedImage.getWidth()-2400, 100);

        getBufferedImage(tesseract,buffered,path,(img,content)->{

            System.out.println("setBuildOn 辉度处理");

            bizLicenseInfo.setBuildOn(content);

        });

    }

    /**

     * 获取营业期限

     * @param tesseract

     * @param bufferedImage

     * @param bizLicenseInfo

     * @param path

     * @throws IOException

     * @throws TesseractException

     */

    private void getBufferedBizLimitImage(Tesseract tesseract, BufferedImage bufferedImage, BizLicenseInfo bizLicenseInfo, String path) throws IOException, TesseractException {

        BufferedImage buffered = ImageFilter.subImage(bufferedImage, 2170, 970, bufferedImage.getWidth()-2400, 100);

        getBufferedImage(tesseract,buffered,path,(img,content)->{

            System.out.println("setBizLimit 辉度处理");

            bizLicenseInfo.setBizLimit(content);

        });

    }

    /**

     * 获取住所

     * @param tesseract

     * @param bufferedImage

     * @param bizLicenseInfo

     * @param path

     * @throws IOException

     * @throws TesseractException

     */

    private void getBufferedAddressImage(Tesseract tesseract, BufferedImage bufferedImage, BizLicenseInfo bizLicenseInfo, String path) throws IOException, TesseractException {

        BufferedImage buffered = ImageFilter.subImage(bufferedImage, 2170, 1070, bufferedImage.getWidth()-2240, 270);

        getBufferedImage(tesseract,buffered,path,(img,content)->{

            System.out.println("setAddress 辉度处理");

            bizLicenseInfo.setAddress(content);

        });

    }

    /**

     * 获取名称

     * @param tesseract

     * @param buffered

     * @param path

     * @param consumer

     */

    private void getBufferedImage(Tesseract tesseract, BufferedImage buffered, String path, BiConsumer<BufferedImage,String> consumer) throws IOException, TesseractException {

        try (OutputStream outputStream = new FileOutputStream(path)) {

//            addressImage = ImageFilter.imageScale(addressImage, ((int) (addressImage.getWidth() * 2.4) + 1), ((int) (addressImage.getHeight() * 2.4) + 1));

            handBrightness(buffered, targetBrightness);

            saveImg(buffered, outputStream);

            tesseract.setLanguage("chi_sim");

            String result = tesseract.doOCR(buffered);

            //留下中文字符、中文标点符号（）【】、

            String regexStr = "[^\\s\\u4e00-\\u9fa5\\(\\)\\uff08\\uff09\\u3001\\u3010\\u3011\\-0-9]+";

            String content = result.replaceAll(regexStr, "")

                    .replaceAll("\\n", "")

                    .replaceAll(" ", "");

            if(consumer!=null){

                consumer.accept(buffered,content);

            }

        }catch (Exception e){

            e.printStackTrace();

            throw e;

        }

    }

    /**

     * 保存图片

     * @param image

     * @param outputStream

     * @throws IOException

     */

    private void saveImg(BufferedImage image,OutputStream outputStream) throws IOException {

        ImageIO.write(image, "jpg", outputStream);

    }

    /**

     * 处理图片辉度

     *

     * @param subImage

     */

    private void handBrightness(BufferedImage subImage, int targetBrightness) {

        int fixedBrightness;

        int birthBrightness = ImageFilter.imageBrightness(subImage);

        System.out.println("brightness = " + birthBrightness);

        fixedBrightness = targetBrightness - birthBrightness;

        //辉度处理

        if (fixedBrightness != 0) {

            subImage = ImageFilter.imageBrightness(subImage, fixedBrightness);

        }

        System.out.println("after brightness = " + ImageFilter.imageBrightness(subImage));

    }

五、解析效果展示

　　5.1、身份证信息识别示例：

　　

　　5.2、营业执照信息识别示例：

　　

参考资料：

https://github.com/firefoxmmx2/IDCardIDentify

https://www.jianshu.com/p/e7915ba6f0e7

https://kefeng.wang/2017/04/22/tess4j/