当前位置:主页 > 移动开发 > Android代码 >

Android动态人脸检测的示例代码(脸数可调)

时间:2021-01-23 10:16:28 | 栏目:Android代码 | 点击:

人脸检测

这里的人脸检测并非人脸识别,但是却可以识别出是否有人,当有人时候,你可以将帧图进行人脸识别(这里推荐Face++的sdk),当然我写的demo中没有加入人脸识别,有兴趣的朋友可以追加。face++

android自带的人脸检测

这里我们用到了人脸检测类为 FaceDetector.这个类提供了强大的人脸检测功能,可以方便我们进行人脸的侦测,因此我们使用他来进行动态的人脸检测,实现原理,其实也挺简单,主要是通过Carmen的回调PreviewCallback 在其中对帧图进行操作,并通过FaceDetector来检测该帧图中是否有人脸。当然如果你想在surfaceview中绘制人脸的范围,可以将画布与其绑定,画完再解绑。

第一步

我们首先来定义一个surfaceview 盖在我们Carmen使用的surfaceview上 进行对人脸范围的绘制

public class FindFaceView extends SurfaceView implements SurfaceHolder.Callback {

  private SurfaceHolder holder;
  private int mWidth;
  private int mHeight;
  private float eyesDistance;

  public FindFaceView(Context context, AttributeSet attrs) {
    super(context, attrs);
    holder = getHolder();
    holder.addCallback(this);
    holder.setFormat(PixelFormat.TRANSPARENT);
    this.setZOrderOnTop(true);
  }

  @Override
  public void surfaceChanged(SurfaceHolder holder, int format, int width,
                int height) {
    mWidth = width;
    mHeight = height;
  }

  @Override
  public void surfaceCreated(SurfaceHolder holder) {

  }

  @Override
  public void surfaceDestroyed(SurfaceHolder holder) {

  }

  public void drawRect(FaceDetector.Face[] faces, int numberOfFaceDetected) {
    Canvas canvas = holder.lockCanvas();
    if (canvas != null) {
      Paint clipPaint = new Paint();
      clipPaint.setAntiAlias(true);
      clipPaint.setStyle(Paint.Style.STROKE);
      clipPaint
          .setXfermode(new PorterDuffXfermode(PorterDuff.Mode.CLEAR));
      canvas.drawPaint(clipPaint);
      canvas.drawColor(getResources().getColor(color.transparent));
      Paint paint = new Paint();
      paint.setAntiAlias(true);
      paint.setColor(Color.GREEN);
      paint.setStyle(Style.STROKE);
      paint.setStrokeWidth(5.0f);
      for (int i = 0; i < numberOfFaceDetected; i++) {
        Face face = faces[i];
        PointF midPoint = new PointF();
        // 获得两眼之间的中间点
        face.getMidPoint(midPoint);
        // 获得两眼之间的距离
        eyesDistance = face.eyesDistance();
        // 换算出预览图片和屏幕显示区域的比例参数
        float scale_x = mWidth / 500;
        float scale_y = mHeight / 600;
        Log.e("eyesDistance=", eyesDistance + "");
        Log.e("midPoint.x=", midPoint.x + "");
        Log.e("midPoint.y=", midPoint.y + "");
        // 因为拍摄的相片跟实际显示的图像是镜像关系,所以在图片上获取的两眼中间点跟手机上显示的是相反方向
        canvas.drawRect((int) (240 - midPoint.x - eyesDistance)
                * scale_x, (int) (midPoint.y * scale_y),
            (int) (240 - midPoint.x + eyesDistance) * scale_x,
            (int) (midPoint.y + 3 * eyesDistance) * scale_y, paint);
      }
      holder.unlockCanvasAndPost(canvas);
    }
  }
}

重要的地方

1. holder = getHolder();获取surfaceholder与我们要绘制人脸范围的画布进行绑定Canvas canvas = holder.lockCanvas();这样我们就可以愉快的进行绘制了,当然前提是我们要拿到人脸的坐标位置。

2. 还有重要的一点,就是要让我们用来盖在Carema上的Surfaceview可以同名,并且设置起在视图树的层级为最高。

 holder.setFormat(PixelFormat.TRANSPARENT);
 this.setZOrderOnTop(true);

第二步

就是我们对人脸进行检测了,当然前提是我们要获得帧图

public class FaceRecognitionDemoActivity extends Activity implements
    OnClickListener {

  private SurfaceView preview;
  private Camera camera;
  private Camera.Parameters parameters;
  private int orientionOfCamera;// 前置摄像头的安装角度
  private int faceNumber;// 识别的人脸数
  private FaceDetector.Face[] faces;
  private FindFaceView mFindFaceView;
  private ImageView iv_photo;
  private Button bt_camera;
  TextView mTV;

  /**
   * Called when the activity is first created.
   */
  @Override
  public void onCreate(Bundle savedInstanceState) {
    super.onCreate(savedInstanceState);
    setContentView(R.layout.main);
  }

  @Override
  protected void onStart() {
    super.onStart();
    iv_photo = (ImageView) findViewById(R.id.iv_photo);
    bt_camera = (Button) findViewById(R.id.bt_camera);
    mTV = (TextView) findViewById(R.id.show_count);
    bt_camera.setOnClickListener(this);

    mFindFaceView = (FindFaceView) findViewById(R.id.my_preview);

    preview = (SurfaceView) findViewById(R.id.preview);
    // 设置缓冲类型(必不可少)
    preview.getHolder().setType(SurfaceHolder.SURFACE_TYPE_PUSH_BUFFERS);
    // 设置surface的分辨率
    preview.getHolder().setFixedSize(176, 144);
    // 设置屏幕常亮(必不可少)
    preview.getHolder().setKeepScreenOn(true);

    preview.getHolder().addCallback(new SurfaceCallback());
  }

  private final class MyPictureCallback implements PictureCallback {

    @Override
    public void onPictureTaken(byte[] data, Camera camera) {
      try {
        Bitmap bitmap = BitmapFactory.decodeByteArray(data, 0,
            data.length);
        Matrix matrix = new Matrix();
        matrix.setRotate(-90);
        Bitmap bmp = Bitmap.createBitmap(bitmap, 0, 0, bitmap
            .getWidth(), bitmap.getHeight(), matrix, true);
        bitmap.recycle();
        iv_photo.setImageBitmap(bmp);
        camera.startPreview();
      } catch (Exception e) {
        e.printStackTrace();
      }
    }

  }

  private final class SurfaceCallback implements Callback {

    @Override
    public void surfaceChanged(SurfaceHolder holder, int format, int width,
                  int height) {
      if (camera != null) {
        parameters = camera.getParameters();
        parameters.setPictureFormat(PixelFormat.JPEG);
        // 设置预览区域的大小
        parameters.setPreviewSize(width, height);
        // 设置每秒钟预览帧数
        parameters.setPreviewFrameRate(20);
        // 设置预览图片的大小
        parameters.setPictureSize(width, height);
        parameters.setJpegQuality(80);
      }
    }

    @Override
    public void surfaceCreated(SurfaceHolder holder) {
      int cameraCount = 0;
      Camera.CameraInfo cameraInfo = new Camera.CameraInfo();
      cameraCount = Camera.getNumberOfCameras();
      //设置相机的参数
      for (int i = 0; i < cameraCount; i++) {
        Camera.getCameraInfo(i, cameraInfo);
        if (cameraInfo.facing == Camera.CameraInfo.CAMERA_FACING_FRONT) {
          try {
            camera = Camera.open(i);
            camera.setPreviewDisplay(holder);
            setCameraDisplayOrientation(i, camera);
            //最重要的设置 帧图的回调
            camera.setPreviewCallback(new MyPreviewCallback());
            camera.startPreview();
          } catch (Exception e) {
            e.printStackTrace();
          }
        }
      }
    }

    @Override
    public void surfaceDestroyed(SurfaceHolder holder) {
    //记得释放,避免OOM和占用
      if (camera != null) {
        camera.setPreviewCallback(null);
        camera.stopPreview();
        camera.release();
        camera = null;
      }
    }

  }

  private class MyPreviewCallback implements PreviewCallback {

    @Override
    public void onPreviewFrame(byte[] data, Camera camera) {
    //这里需要注意,回调出来的data不是我们直接意义上的RGB图 而是YUV图,因此我们需要
    //将YUV转化为bitmap再进行相应的人脸检测,同时注意必须使用RGB_565,才能进行人脸检测,其余无效
      Camera.Size size = camera.getParameters().getPreviewSize();
      YuvImage yuvImage = new YuvImage(data, ImageFormat.NV21,
          size.width, size.height, null);
      ByteArrayOutputStream baos = new ByteArrayOutputStream();
      yuvImage.compressToJpeg(new Rect(0, 0, size.width, size.height),
          80, baos);
      byte[] byteArray = baos.toByteArray();
      detectionFaces(byteArray);
    }
  }

  /**
   * 检测人脸
   *
   * @param data 预览的图像数据
   */
  private void detectionFaces(byte[] data) {
    BitmapFactory.Options options = new BitmapFactory.Options();
    Bitmap bitmap1 = BitmapFactory.decodeByteArray(data, 0, data.length,
        options);
    int width = bitmap1.getWidth();
    int height = bitmap1.getHeight();
    Matrix matrix = new Matrix();
    Bitmap bitmap2 = null;
    FaceDetector detector = null;
    //设置各个角度的相机,这样我们的检测效果才是最好
    switch (orientionOfCamera) {
      case 0:
        //初始化人脸检测(下同)
        detector = new FaceDetector(width, height, 10);
        matrix.postRotate(0.0f, width / 2, height / 2);
        // 以指定的宽度和高度创建一张可变的bitmap(图片格式必须是RGB_565,不然检测不到人脸)
        bitmap2 = Bitmap.createBitmap(width, height, Bitmap.Config.RGB_565);
        break;
      case 90:
        detector = new FaceDetector(height, width, 1);
        matrix.postRotate(-270.0f, height / 2, width / 2);
        bitmap2 = Bitmap.createBitmap(height, width, Bitmap.Config.RGB_565);
        break;
      case 180:
        detector = new FaceDetector(width, height, 1);
        matrix.postRotate(-180.0f, width / 2, height / 2);
        bitmap2 = Bitmap.createBitmap(width, height, Bitmap.Config.RGB_565);
        break;
      case 270:
        detector = new FaceDetector(height, width, 1);
        matrix.postRotate(-90.0f, height / 2, width / 2);
        bitmap2 = Bitmap.createBitmap(height, width, Bitmap.Config.RGB_565);
        break;
    }
    //设置支持的面数(最大支持检测多少人的脸 ,可以根据需要调整,不过需要与findFaces中的参数数值相同,否则会抛出异常)
    faces = new FaceDetector.Face[10];
    Paint paint = new Paint();
    paint.setDither(true);
    Canvas canvas = new Canvas();
    canvas.setBitmap(bitmap2);
    canvas.setMatrix(matrix);
    // 将bitmap1画到bitmap2上(这里的偏移参数根据实际情况可能要修改)
    canvas.drawBitmap(bitmap1, 0, 0, paint);
    //这里通过向findFaces中传递帧图转化后的bitmap和最大检测的人脸数face,返回检测后的人脸数
    faceNumber = detector.findFaces(bitmap2, faces);
    mTV.setText("facnumber----" + faceNumber);
    mTV.setTextColor(Color.RED);
    //这里就是我们的人脸识别,绘制识别后的人脸区域的类
    if (faceNumber != 0) {
      mFindFaceView.setVisibility(View.VISIBLE);
      mFindFaceView.drawRect(faces, faceNumber);
    } else {
      mFindFaceView.setVisibility(View.GONE);
    }
    bitmap2.recycle();
    bitmap1.recycle();
  }

  /**
   * 设置相机的显示方向(这里必须这么设置,不然检测不到人脸)
   *
   * @param cameraId 相机ID(0是后置摄像头,1是前置摄像头)
   * @param camera  相机对象
   */
  private void setCameraDisplayOrientation(int cameraId, Camera camera) {
    Camera.CameraInfo info = new Camera.CameraInfo();
    Camera.getCameraInfo(cameraId, info);
    int rotation = getWindowManager().getDefaultDisplay().getRotation();
    int degree = 0;
    switch (rotation) {
      case Surface.ROTATION_0:
        degree = 0;
        break;
      case Surface.ROTATION_90:
        degree = 90;
        break;
      case Surface.ROTATION_180:
        degree = 180;
        break;
      case Surface.ROTATION_270:
        degree = 270;
        break;
    }

    orientionOfCamera = info.orientation;
    int result;
    if (info.facing == Camera.CameraInfo.CAMERA_FACING_FRONT) {
      result = (info.orientation + degree) % 360;
      result = (360 - result) % 360;
    } else {
      result = (info.orientation - degree + 360) % 360;
    }
    camera.setDisplayOrientation(result);
  }

  @Override
  public void onClick(View v) {
    switch (v.getId()) {
      case R.id.bt_camera:
        if (camera != null) {
          try {
            camera.takePicture(null, null, new MyPictureCallback());
          } catch (Exception e) {
            e.printStackTrace();
          }
        }
        break;
    }
  }
}

到这里我们的人脸识别就已经大功告成。demo地址

如果您想了解更多关于人脸识别方面的只是,先去关注并了解OpenCV。

您可能感兴趣的文章:

相关文章