golang 通过docker 搭建 ocr识别

开源gosseract

地址 gosseract

1. 首先下载docker 和docker-compose ，具体的查看官网
注意 docker版本和docker-compose 版本一致

2. gosseract 在源码中找到dockerfile 文件如下(做了稍微的修改，只安装golang 和 tesseract-ocr/)：

##### docker build -f docker_file -t otiai10:0.1 .
# This is a working example of setting up tesseract/gosseract,
# and also works as an example runtime to use gosseract package.
# You can just hit `docker run -it --rm otiai10/gosseract`
# to try and check it out!

#####
FROM golang:latest
LABEL maintainer="Hiromu Ochiai <otiai10@gmail.com>"

RUN apt-get update -qq

# You need librariy files and headers of tesseract and leptonica.
# When you miss these or LD_LIBRARY_PATH is not set to them,
# you would face an error: "tesseract/baseapi.h: No such file or directory"
RUN apt-get install -y -qq libtesseract-dev libleptonica-dev

RUN apt update \
    && apt install -y \
      ca-certificates \
      libtesseract-dev=4.1.1-2+b1 \
      tesseract-ocr=4.1.1-2+b1

# In case you face TESSDATA_PREFIX error, you minght need to set env vars
# to specify the directory where "tessdata" is located.
ENV TESSDATA_PREFIX=/usr/share/tesseract-ocr/4.00/tessdata/

# Load languages.
# These {lang}.traineddata would b located under ${TESSDATA_PREFIX}/tessdata.
# 安装中文
RUN apt-get install -y -qq \
  tesseract-ocr-eng \
  tesseract-ocr-deu \
  #tesseract-ocr-chi_sim \
  tesseract-ocr-jpn
 #检查「tesseract」支持的语言
 # tesseract --list-langs
# See https://github.com/tesseract-ocr/tessdata for the list of available languages.
# https://github.com/tesseract-ocr/tessdata/tree/4.00 或者下载 https://tesseract-ocr.github.io/tessdoc/Data-Files
# If you want to download these traineddata via `wget`, don't forget to locate
# downloaded traineddata under ${TESSDATA_PREFIX}/tessdata.


RUN go env -w GO111MODULE=on
RUN  go env -w GOPROXY=https://goproxy.cn,direct
#RUN go get -t github.com/otiai10/gosseract

RUN /go/src
#RUN cd ${GOPATH}/src/github.com/otiai10/gosseract && go test

# Now, you've got complete environment to play with "gosseract"!
# For other OS, check https://github.com/otiai10/gosseract/tree/main/test/runtimes

3. 在Dockerfile 目录下面编译镜像文件,命令如下(注意如果文件不是Dockerfile 需要 -f xxxxfile 指定编排文件)

docker build  -t orc .

4. 下载过来查看镜像

docker images 

5. 从官网下载源码运行,新建 docker-compose.yml 文件

创建网络

 docker network create go_app_orc

编排 挂在你的项目目录 orc_test 挂在到容器里面 /go/src/orc_test

version: '3'
# docker network create go_app_orc
# docker cp /Users/gitxuzan/Downloads/chi_sim.traineddata orc:/usr/share/tesseract-ocr/4.00/tessdata
#  docker exec -it orc_v2 bash
services:
  golang:

    image: orc
    container_name: orc
    networks:
      - web
    volumes:
      - "../orc_test:/go/src/orc_test"

    tty: true
    environment:
      TZ: "Asia/Shanghai"

networks:
  web:
    external:
      name: go_app_orc

6. 启动容器

docker-compose up -d

7. 进入容器

docker exec -it orc bash

8. 进入挂载项目运行

代码如下:

package main

import (
	"fmt"
	"github.com/otiai10/gosseract/v2"
	"log"
)

func main() {
	client := gosseract.NewClient()
	defer client.Close()

	//client.SetLanguage("eng","chi_sim")
	//client.SetLanguage("chi_sim")
	


	client.SetImage("test2.png")
	text, _ := client.Text()
	fmt.Println(text)
	// Hello, World!
}

我用的go mod 管理的 go mod tidy , go run main.go,识别结果如下

9. 最后如果需要识别中文，需要下载中文包拷贝到容器
中文包下载地址1
中文包下载地址2

拷贝到容器
docker cp /Users/gitxuzan/Downloads/chi_sim.traineddata orc:/usr/share/tesseract-ocr/4.00/tessdata
# 查看是否有中文包
tesseract --list-langs

设置识别语言

```go
package main

import (
	"fmt"
	"github.com/otiai10/gosseract/v2"
	"log"
)

func main() {
	client := gosseract.NewClient()
	defer client.Close()

	client.SetLanguage("eng","chi_sim")
	
	client.SetImage("test2.png")
	text, _ := client.Text()
	fmt.Println(text)
	// Hello, World!
}

识别中文结果:

现在存在的问题¹

更新 4.1.1镜像
给作者提出issues 反馈

FROM debian:bullseye-slim

RUN apt-get update -qq
RUN apt-get install -y \
  git \
  golang \
  libtesseract-dev=4.1.1-2.1 \
  tesseract-ocr-eng

ENV GOPATH=/root/go

ADD . ${GOPATH}/src/github.com/otiai10/gosseract
WORKDIR ${GOPATH}/src/github.com/otiai10/gosseract

RUN tesseract --version

CMD ["go", "test", "-v", "./..."]

另外一种方式，不用cgo ，使用原始命令调用

下载地址: https://sourceforge.net/projects/tesseract-ocr.mirror/

windows 界面化下一步下一步安装，需要选择下中文包的下载

例如我这边安装目录是 C:\Program Files\Tesseract-OCR

然后吧这个目录添加到系统Path 里面
验证:
tesseract --version

下载语言包

# 简体中文
curl -O https://github.com/tesseract-ocr/tessdata/raw/main/chi_sim.traineddata
# 繁体中文
curl -O https://github.com/tesseract-ocr/tessdata/raw/main/chi_tra.traineddata
# 英文
curl -O https://github.com/tesseract-ocr/tessdata/raw/main/eng.traineddata

下载好的语言包放到对应的位置

例如mac上

brew --prefix tesseract
/opt/homebrew/opt/tesseract/share/tessdata

windows
C:\Program Files\Tesseract-OCR\tessdata

测试

使用英文和中文
tesseract example.png stdout -l chi_sim+eng 输出到标注输出
tesseract example.png output -l chi_sim+eng 输入到文件里面

写个程序调用

func main() {
	// 设置输入图片和语言参数
	inputImage := "example.png"
	lang := "chi_sim+eng"

	// 根据操作系统设置命令
	var cmd *exec.Cmd
	if runtime.GOOS == "windows" {
		cmd = exec.Command("tesseract.exe", inputImage, "stdout", "-l", lang)
	} else {
		cmd = exec.Command("tesseract", inputImage, "stdout", "-l", lang)
	}

	// 创建一个缓冲区来捕获命令输出
	var outBuffer, errBuffer bytes.Buffer
	cmd.Stdout = &outBuffer
	cmd.Stderr = &errBuffer

	// 运行命令
	if err := cmd.Run(); err != nil {
		log.Println("Error executing command: %v\n", err)
		return
	}

	// 打印标准输出
	fmt.Println(outBuffer.String())
}

这个太复杂的识别有点不准

---

1. 4.0.0 不能设置白名单,需要升级成4.1.1 ↩︎