>>分享Android开发相关的技术 书籍支持  卫琴直播  品书摘要  在线测试  资源下载  联系我们
发表一个新主题 开启一个新投票 回复文章 您是本文章第 22892 个阅读者 刷新本主题
 * 贴子主题:  Android Camera2.0 API实现摄像头预览并获取人脸关键坐标 回复文章 点赞(0)  收藏  
作者:sunshine    发表时间:2020-03-08 13:20:07     消息  查看  搜索  好友  邮件  复制  引用

                                                                                                

Android Camera2.0 API实现摄像头预览并获取人脸关键坐标

Android 5.0(API Level 21)以后推出了新的camera2.0 API,原有的Camera1.0已被废弃,确实新的camera API有更好的架构,更低的耦合,可以使开发人员发挥更大的空间。

         API简介

主要的类有以下几个:

1.CameraManager :所有camera的管理类,可以通过调用getSystemService()得到其实例,其中的方法getCameraCharacteristics()可以获取代表camera特征的类CameraCharacteristics的实例,该特征类中封装了摄像头的各种属性参数,比如是前置摄像头还是后置摄像头等等。

2.CameraDevice:代表一个摄像头,可以通过其方法createCaptureSession()和  createCaptureRequest()创建CameraCaptureSession以及CaptureRequest的对象实例。

3.CameraDevice.StateCallback:CameraDevice内部类,该类用于接收相机的连接状态的更新。比如当相机打开成功后会回调其中的onOpened方法,当相机连接断开时会回调其中的onDisconnected方法。

4.CameraCaptureSession:代表一次拍摄会话,通过setRepeatingRequest()可以开启摄像头预览,capture()方法可以拍照,还有两个内部类CameraCaptureSession.StateCallback以及CaptureCallback,和CamearDevice的StateCallback一样,可以监听预览或拍摄的过程中出现的一些情况。

         人脸检测

主要使用android.hardware.camera2.params.Face,这是Camera2.0自带的一个类,可以在createCaptureSession()中从CaptureResult得到,该Face类中封装了代表人脸基本位置的矩形框,是一个Rect对象,其他还能返回的有两眼和嘴巴的位置,分别都是Point数组。

         主要代码

我们使用TextureView作为摄像头预览输出的载体,创建一个类实现TextureView.SurfaceTextureListener接口,在重写的方法onSurfaceTextureAvailable()中开启摄像头        

       @Override
     public  void  onSurfaceTextureAvailable(SurfaceTexture surface,  int width,  int height) {

         //设置Camera初始参数
        setUpCamera();

         //获取Surface表面
        surfaceTexture = surface ;

         //设置SurfaceTexture默认大小
        surfaceTexture.setDefaultBufferSize(mPreviewSize.getWidth(),mPreviewSize.getHeight());

         //开启后台线程
        openBackgroundThread();

         //开启相机
        openCamera();

    }

     一、首先是setUpCamera()函数设置摄像头的初始化参数,包括人脸检测的开启        

  /**
     * 设置camera2.0的初始化参数
     */

     private  void  setUpCamera() {
        cameraManager = (CameraManager)mContext.getSystemService(Context.CAMERA_SERVICE);
         try{
             for (String id : cameraManager.getCameraIdList()) {

                 //获取代表摄像头特征类characteristics
                characteristics = cameraManager.getCameraCharacteristics(id);

                 //如果是前置摄像头
                 if (characteristics.get(CameraCharacteristics.LENS_FACING) == CameraCharacteristics.LENS_FACING_FRONT) {
                    mCameraId = id ;

                    StreamConfigurationMap streamConfigurationMap = characteristics.get(CameraCharacteristics.SCALER_STREAM_CONFIGURATION_MAP);

                    sizes = streamConfigurationMap.getOutputSizes(SurfaceHolder.class);

                     //设置预览大小
                    mPreviewSize = sizes[ 0];

                     //获取人脸检测参数
                     int[] FD =characteristics.get(CameraCharacteristics.STATISTICS_INFO_AVAILABLE_FACE_DETECT_MODES);
                     int maxFD=characteristics.get(CameraCharacteristics.STATISTICS_INFO_MAX_FACE_COUNT);

                     if (FD.length> 0) {
                        List<Integer> fdList =  new ArrayList<>();
                         for ( int FaceD : FD
                                ) {
                            fdList.add(FaceD);
                            Log.e(TAG,  "setUpCameraOutputs: FD type:" + Integer.toString(FaceD));
                        }
                        Log.e(TAG,  "setUpCameraOutputs: FD count" + Integer.toString(maxFD));

                         if (maxFD >  0) {
                            mFaceDetectSupported =  true;
                            mFaceDetectMode = Collections.max(fdList);
                        }
                    }
                }
            }
        }  catch ( CameraAccessException e ){
            e.printStackTrace();
        }
    }

     二、然后openCamera()方法开启摄像头,在其中检查是否开启摄像头权限        

    /**
     * 查看摄像头并开启摄像机
     */

     public  void  openCamera(){

         try {

             //判断是否开启摄像头权限
             if (PermissionChecker.checkSelfPermission( mContext , Manifest.permission.CAMERA) == PackageManager.PERMISSION_GRANTED) {
                cameraManager.openCamera(mCameraId, cameraCallback, mBackgroundHandler);
            } else {
                Toast.makeText( mContext , "请打开摄像头权限",Toast.LENGTH_SHORT).show();
            }

        }  catch (CameraAccessException e) {
            e.printStackTrace();
        }
    }

     三、摄像头开启成功后,在CameraDevice.StateCallback的onOpened()方法中开启预览                

       //cameraCallback回调接口
     private CameraDevice.StateCallback cameraCallback =  new CameraDevice.StateCallback(){

         //若摄像机打开成功则回调此方法
         @Override
         public  void  onOpened(CameraDevice camera) {
             //获取cameraDevice
            cameraDevice = camera;
             //开启预览
            startPreview();
        }

         //摄像机连接断开回调此方法
         @Override
         public  void  onDisconnected(CameraDevice camera) {

             if(cameraDevice !=  null ){
                cameraDevice.close();
            }
        }
         //出现异常回调此方法
         @Override
         public  void  onError(CameraDevice camera,  int error) {
             if(cameraDevice !=  null ){
                cameraDevice.close();
            }
        }

    };

     四、startPreview()方法开启预览,并打印出返回的人脸位置坐标        

        public  void  startPreview(){
         try{
            Surface surface =  new Surface(surfaceTexture);
            previewRequestBuilder = cameraDevice.createCaptureRequest(CameraDevice.TEMPLATE_PREVIEW);
            previewRequestBuilder.addTarget(surface);
             /*previewRequestBuilder.addTarget(mImageReader.getSurface());*/
            previewRequestBuilder.set(CaptureRequest.STATISTICS_FACE_DETECT_MODE,
                    CameraMetadata.STATISTICS_FACE_DETECT_MODE_FULL);
            cameraDevice.createCaptureSession(Arrays.asList(surface,mImageReader.getSurface()),  new CameraCaptureSession.StateCallback() {
                 @Override
                 public  void  onConfigured(@NonNull CameraCaptureSession session) {
                     try {
                         //构建captureRequest对象
                        captureRequest = previewRequestBuilder.build();
                         //设置人脸检测
                        setFaceDetect(previewRequestBuilder,mFaceDetectMode);
                        captureSession = session;
                        captureSession.setRepeatingRequest(captureRequest,  new CameraCaptureSession.CaptureCallback() {

                             /**
                             * 对摄像头返回的结果进行处理,并获取人脸数据
                             *  @param result 摄像头数据
                             */

                             private  void  process(CaptureResult result) {

                                 //获得Face类
                                Face face[]=result.get(CaptureResult.STATISTICS_FACES);

                                 //如果有人脸的话
                                 if (face.length> 0 ){
                                    Log.e(TAG,  "face detected " + Integer.toString(face.length));

                                     //获取人脸矩形框
                                    Rect bounds = face[ 0].getBounds();

                                     float y = mPreviewSize.getHeight()/ 2 - bounds.top ;

                                    Log.e( "height" , String.valueOf(mPreviewSize.getWidth()));
                                    Log.e( "top" , String.valueOf(y));
                                    Log.e( "left" ,  String.valueOf(bounds.left));
                                    Log.e( "right" , String.valueOf(bounds.right));

                                    }

                                }

                             @Override
                             public  void  onCaptureStarted(CameraCaptureSession session, CaptureRequest request,  long timestamp,  long frameNumber) {
                                 super.onCaptureStarted(session, request, timestamp, frameNumber);
                            }

                             @Override
                             public  void  onCaptureProgressed(CameraCaptureSession session, CaptureRequest request, CaptureResult partialResult) {
                                process(partialResult);
                            }

                             @Override
                             public  void  onCaptureCompleted(CameraCaptureSession session, CaptureRequest request, TotalCaptureResult result) {
                                process(result);
                            }
                        },mBackgroundHandler);

                        }  catch (CameraAccessException e) {
                        e.printStackTrace();
                    }
                }

                 @Override
                 public  void  onConfigureFailed(CameraCaptureSession session) {

                }

                },  null);
        } catch (CameraAccessException e){
            e.printStackTrace();
        }

    }

     最后在布局文件中添加一个TextureView,并在Activity中获取其视图,为其设置SurfaceTextureListener,也就是刚才自定义的类,就OK了。
                                    
                                                                    
----------------------------
原文链接:https://blog.csdn.net/SakuraMashiro/article/details/78334248

程序猿的技术大观园:www.javathinker.net



[这个贴子最后由 flybird 在 2020-03-09 22:53:27 重新编辑]
  Java面向对象编程-->对象的生命周期
  JavaWeb开发-->JSP技术详解(Ⅰ)
  JSP与Hibernate开发-->JPA API的高级用法
  Java网络编程-->XML数据处理
  精通Spring-->虚拟DOM和render()函数
  Vue3开发-->虚拟DOM和render()函数
  Android自动化测试之Robotium学习
  Android网络编程之WebKit应用
  Android ListView高度问题
  浅析Android的通知系统
  Android JSON网络数据交换
  Android 滚动Tab
  Android代码混淆的实践
  Android 抽屉导航
  Android开发实践:Android.mk模板
  Android--Widget开发
  Roboletric+Retrofit2单元测试
  Android OpenGL ES2.0学习:基础三角形案例
  Android性能优化技巧
  UI渲染机制以及优化
  【Android 修炼手册】Gradle 篇 -- Gradle 的基本使用
  更多...
 IPIP: 已设置保密
楼主      
1页 0条记录 当前第1
发表一个新主题 开启一个新投票 回复文章


中文版权所有: JavaThinker技术网站 Copyright 2016-2026 沪ICP备16029593号-2
荟萃Java程序员智慧的结晶,分享交流Java前沿技术。  联系我们
如有技术文章涉及侵权,请与本站管理员联系。