OpenAlpr车牌识别功能实现

OpenAlpr车牌识别功能实现

关于OpenALPR

OpenALPR是一种使用C ++编写的开源自动车牌识别库，支持多个国家多个地区的车牌。而最近公司想做车牌识别这一块业务，要支持全球多个国家的车牌，于是写了个Demo测试OpenAlpr的接口（有2000次免费机会）。

功能实现

import android.app.Activity;
t.ContentValues;
t.Intent;
aphics.Bitmap;
aphics.BitmapFactory;
import android.Uri;
import android.os.Build;
import android.os.Bundle;
import android.os.Environment;
import android.provider.MediaStore;
import android.support.v7.app.AppCompatActivity;
format.DateFormat;
import android.view.View;
import android.widget.AdapterView;
import android.widget.ArrayAdapter;
import android.widget.Button;
import android.widget.ImageView;
import android.widget.Spinner;
import android.widget.TextView;
import android.widget.Toast;import com.openalpr.api.DefaultApi;
import com.openalpr.api.invoker.ApiException;
import com.dels.InlineResponse200;
import com.yalantis.ucrop.UCrop;
import com.yalantis.ucrop.UCropActivity;import java.io.File;
import java.io.FileInputStream;
import java.io.IOException;
import java.util.Calendar;
import java.util.Locale;public class MainActivity extends AppCompatActivity {private final DefaultApi api = new DefaultApi();private static final String TAG = "MainActivity";Button reconize,select;TextView result,plate;ImageView showImage;Spinner spinner;String secretKey = "sk_960d6f4c62d1c20452c7613e";String url = ".jpg?imageView2/2/w/752|watermark/2/text/TEkzMw0K5rG96L2m5LmL5a62/font/5b6u6L2v6ZuF6buR/fontsize/270/fill/d2hpdGU=/dissolve/100/gravity/SouthEast/dx/5/dy/5";File cameraFile,cropFile;private boolean isFromfile = false;String country = "us";final Integer recognizeVehicle = 0;final String state = "";final Integer returnImage = 1;final Integer topn = 10;final String prewarp = "";private String cropPath = "";String[] countrys ;String[] countryCodes ;@Overrideprotected void onCreate(Bundle savedInstanceState) {Create(savedInstanceState);setContentView(R.layout.activity_main);reconize = (Button) findViewById(ize);select = (Button) findViewById(R.id.selectImage);result = (TextView) findViewById(sult);showImage = (ImageView) findViewById(R.id.imageShow);plate = (TextView) findViewById(R.id.plate);spinner = (Spinner) findViewById(R.id.spinner);
//        GlideUtils.setUrlImage(this,url,showImage);countrys = getResources().getStringArray(untrys);countryCodes = getResources().getStringArray(untry_codes);select.setOnClickListener(new View.OnClickListener() {@Overridepublic void onClick(View v) {String filename = DateFormat.format("yyyyMMdd_hhmmss", Instance(Locale.CHINA)) + ".jpg";cameraFile = new ExternalStorageDirectory() + "/Pictures/Camera/" + filename);if(!ists()){try {ateNewFile();} catch (IOException e) {e.printStackTrace();}}Intent intent0 = new Intent(MediaStore.ACTION_IMAGE_CAPTURE);if (Build.VERSION.SDK_INT < 24) {intent0 = new Intent(MediaStore.ACTION_IMAGE_CAPTURE);intent0.putExtra("as.CAMERA_FACING", 0);intent0.putExtra(MediaStore.EXTRA_OUTPUT, Uri.fromFile(cameraFile));} else {ContentValues contentValues = new ContentValues(1);intent0.putExtra("as.CAMERA_FACING", 0);contentValues.put(MediaStore.Images.Media.DATA, AbsolutePath());Uri uri = ContentResolver().insert(MediaStore.Images.Media.EXTERNAL_CONTENT_URI, contentValues);intent0.putExtra(MediaStore.EXTRA_OUTPUT, uri);}isFromfile = true;startActivityForResult(intent0, 200);}});reconize.setOnClickListener(new View.OnClickListener() {@Overridepublic void onClick(View view) {result.setText("正在识别。。。。。");plate.setText("");try {new Thread(){@Overridepublic void run() {super.run();InlineResponse200 response = null;try {if(isFromfile){response = izeFile(cropFile, secretKey, country, recognizeVehicle, state, returnImage, topn, prewarp);}else {response = izeUrl(url, secretKey, country, recognizeVehicle, state, returnImage, topn, prewarp);}} catch (ApiException e) {e.printStackTrace();}final InlineResponse200 finalResponse = response;runOnUiThread(new Runnable() {@Overridepublic void run() {if(finalResponse == null){Toast.makeText(MainActivity.this, "识别失败", Toast.LENGTH_SHORT).show();return;}Toast.makeText(MainActivity.this, "识别成功", Toast.LENGTH_SHORT).show();Results()!=null && Results().size()>=1)plate.setText("车牌号："&#Results().get(0).getPlate());result.setText("返回结果:"&#String());}});}}.start();} catch (Exception e) {e.printStackTrace();}}});ArrayAdapter<String> product_adapter = new ArrayAdapter<String>(this, android.R.layout.select_dialog_singlechoice, countrys);spinner.setAdapter(product_adapter);spinner.setSelection(12);spinner.setOnItemSelectedListener(new AdapterView.OnItemSelectedListener() {@Overridepublic void onItemSelected(AdapterView<?> parent, View view, int position, long id) {country = countryCodes[position];}@Overridepublic void onNothingSelected(AdapterView<?> parent) {}});}@Overrideprotected void onActivityResult(int requestCode, int resultCode, Intent data) {ActivityResult(requestCode, resultCode, data);if(resultCode != RESULT_OK){return;}switch (requestCode) {case 200:if (!ists()) {return ;}Bitmap bitmap = null;if (resultCode== Activity.RESULT_OK){try {FileInputStream fis = new FileInputStream(cameraFile);bitmap = BitmapFactory.decodeStream(fis);fis.close();} catch (Exception e) {e.printStackTrace();}if (bitmap!=null) {showImage.setImageBitmap(bitmap);}cropPath = startUCrop(Path(),UCrop.REQUEST_CROP,16,9);}break;case UCrop.REQUEST_CROP:final Uri resultUri = Output(data);showImage.setImageURI(resultUri);cropFile = new File(cropPath);break;default:break;}}/*** 启动裁剪* @param activity 上下文* @param sourceFilePath 需要裁剪图片的绝对路径* @param requestCode 比如：UCrop.REQUEST_CROP* @param aspectRatioX 裁剪图片宽高比* @param aspectRatioY 裁剪图片宽高比* @return*/public static String startUCrop(Activity activity, String sourceFilePath,int requestCode, float aspectRatioX, float aspectRatioY) {Uri sourceUri = Uri.fromFile(new File(sourceFilePath));File outDir = ExternalStoragePublicDirectory(Environment.DIRECTORY_PICTURES);if (!ists()) {outDir.mkdirs();}File outFile = new File(outDir, System.currentTimeMillis() + ".jpg");//裁剪后图片的绝对路径String cameraScalePath = AbsolutePath();Uri destinationUri = Uri.fromFile(outFile);//初始化，第一个参数：需要裁剪的图片；第二个参数：裁剪后图片UCrop uCrop = UCrop.of(sourceUri, destinationUri);//初始化UCrop配置UCrop.Options options = new UCrop.Options();//设置裁剪图片可操作的手势options.setAllowedGestures(UCropActivity.SCALE, UCropActivity.ROTATE, UCropActivity.ALL);//是否隐藏底部容器，默认显示options.setHideBottomControls(true);
//        //设置toolbar颜色
//        options.Color(activity, lorPrimary));
//        //设置状态栏颜色
//        options.Color(activity, lorPrimary));//是否能调整裁剪框options.setFreeStyleCropEnabled(true);//UCrop配置uCrop.withOptions(options);//设置裁剪图片的宽高比，比如16：9uCrop.withAspectRatio(aspectRatioX, aspectRatioY);//uCrop.useSourceImageAspectRatio();//跳转裁剪页面uCrop.start(activity, requestCode);return cameraScalePath;}}

主要分为拍照，裁剪（主要是为了区域更小识别更准确），选取区域，上传图片，获取解析结果。

测试结果

识别速度这一块与图片体积有关，识别准确率主要跟地区选取和图像清晰度这一块有关，总体来说功能非常强大，支持多个地区，而且识别率还非常高，支持的地区主要有

        <item>阿根廷</item><item>澳大利亚</item><item>巴西</item><item>中国</item><item>欧洲</item><item>英国</item><item>印度</item><item>印度尼西亚</item><item>日本</item><item>韩国</item><item>中东</item><item>新西兰</item><item>北美</item><item>俄罗斯</item><item>沙特阿拉伯</item><item>南非</item><item>泰国</item><item>阿拉伯联合酋长国</item>对应的code值为<item>ar</item><item>au</item><item>br</item><item>cn</item><item>eu</item><item>en</item><item>in</item><item>id</item><item>jp</item><item>kr</item><item>me</item><item>nz</item><item>us</item><item>ru</item><item>sa</item><item>za</item><item>th</item><item>ae</item>

代码地址
GitHub地址