android 离线文字识别开源库 tesseract

前言

• 离线文字识别开源库，本文只介绍如何使用
• tess-two github地址: https://github.com/rmtheis/tess-two

导入依赖

implementation 'com.rmtheis:tess-two:9.1.0'

下载导入训练数据

• 下载地址https://github.com/tesseract-ocr/tessdata
• 大多数情况只需要识别简体中文和英文对应数据包：chi_sim.traineddata，eng.traineddata
• 项目中创建assets把对应的数据包导入

调用

object TessUtil {
	// 指定识别的语言英文加中文
    private const val LANGUAGE = "eng+chi_sim"
	// 指定数据包在设备中路径 
    private val mDataPath = PathUtils.getExternalAppCachePath() + File.separator + "tesseract" + File.separator

    /**
     * @param imagePath 识别的图片地址
     * @param callback 结果回调
     */
    fun textRecognition(imagePath: String, callback: Callback) {
        textRecognition(ImageUtils.getBitmap(imagePath), callback)
    }

    /**
     * @param bitmap 识别的图片
     * @param callback 结果回调
     */
    fun textRecognition(bitmap: Bitmap, callback: Callback) {
        ThreadUtils.getSinglePool()
            .execute {
                try {
                    // 复制字体库到手机sd卡中
                    val path = "${mDataPath}tessdata"
                    copyFont(path, "chi_sim.traineddata")
                    copyFont(path, "eng.traineddata")
                    // 开启文字识别
                    val tessBaseAPI = TessBaseAPI()
                    tessBaseAPI.init(mDataPath, LANGUAGE)
                    tessBaseAPI.setImage(ImageUtils.toGray(bitmap))
                    val utF8Text = tessBaseAPI.utF8Text
                    ThreadUtils.runOnUiThread {
                        callback.success(utF8Text)
                    }
                    tessBaseAPI.clear()
                } catch (e: Exception) {
                    ThreadUtils.runOnUiThread {
                        callback.failure(e)
                    }
                }
            }
    }

    private fun copyFont(path: String, fontName: String) {
        FileUtils.createOrExistsDir(path)
        val fontPath = path + File.separator + fontName
        if (!FileUtils.isFileExists(fontPath)) {
            ResourceUtils.copyFileFromAssets(fontName, fontPath)
        }
    }

    interface Callback {

        fun success(result: String)

        fun failure(e: Exception) {
            e.printStackTrace()
        }
    }
}

更新

• 作者将库迁移到Tesseract4Android了，原来的也可以用，更新一下

修改依赖

• 选择其中一个导入，OpenMP简单理解，就是对多核CPU机器进行了优化，测试OpenMP识别速度更快一些。

    // 使用标准
    implementation 'cz.adaptech.tesseract4android:tesseract4android:4.1.1a'
    // 使用OpenMP
    implementation 'cz.adaptech.tesseract4android:tesseract4android-openmp:4.1.1a'

修改调用

object TessOCRUtils {

    private const val LANGUAGE = "eng+chi_sim"

    //    private const val LANGUAGE = "chi_sim"
    //  /storage/emulated/0/Android/data/package/cache/tesseract/
    private val DATA_PATH = PathUtils.getExternalAppCachePath() + File.separator + "tesseract" + File.separator

    /**
     * @param imagePath 识别的图片地址
     * @param callback 结果回调
     */
    fun textRecognition(imagePath: String, callback: Callback) {
        textRecognition(ImageUtils.getBitmap(imagePath), callback)
    }

    /**
     * @param bitmap 识别的图片
     * @param callback 结果回调
     */
    fun textRecognition(bitmap: Bitmap, callback: Callback) {
        ThreadUtils.getSinglePool()
            .execute {
                try {
                    // 复制字体库到手机sd卡中
                    val path = "${DATA_PATH}tessdata"
                    copyFont(path, "chi_sim.traineddata")
                    copyFont(path, "eng.traineddata")
                    // 开启文字识别
                    val tessBaseAPI = TessBaseAPI()
                    if (!tessBaseAPI.init(DATA_PATH, LANGUAGE)) {
                        tessBaseAPI.recycle()
                        throw RuntimeException("tess初始化失败")
                    } else {
                        // 图片灰度处理，提高识别率
                        val gray = ImageUtils.toGray(bitmap)
                        tessBaseAPI.setImage(gray)
                        val utF8Text = tessBaseAPI.utF8Text
                        ThreadUtils.runOnUiThread {
                            callback.success(utF8Text)
                        }
                        tessBaseAPI.recycle()
                    }
                } catch (e: Exception) {
                    ThreadUtils.runOnUiThread {
                        callback.failure(e)
                    }
                }
            }
    }

    private fun copyFont(path: String, fontName: String) {
        FileUtils.createOrExistsDir(path)
        val fontPath = path + File.separator + fontName
        if (!FileUtils.isFileExists(fontPath)) {
            ResourceUtils.copyFileFromAssets(fontName, fontPath)
        }
    }

    interface Callback {

        fun success(result: String)

        fun failure(e: Exception) {
            e.printStackTrace()
        }
    }
}