CameraX 封装二维码扫描组件

逮虾户

发布于 2024-01-27 10:11:38

2160

发布于 2024-01-27 10:11:38

文章被收录于专栏：逮虾户

简介

cameraX已经出来有一段时间了，现在已经从alpha版本到现在的beta3版本。其中内部的代码版本跨度特别大，而且资料相对来说只有官方的demo比较可以参考，所以最近完成了项目的开发之后，把经验分享一下提供给各位。

仓库地址

究极花里胡哨camreaX二维码扫描

二维码扫描小优化

去除zxing额外支持的格式（有争议的点，其实并没有特别大的差距）

MultiFormatReader的decodeWithState()是使用方的入口方法，内部调用了decodeInternal()，输入是相机的一帧数据，如果抛了NotFoundException，则表示没找到二维码；如果返回了Result，则表示找到了二维码，并解析完成。其中，readers变量是一个数组，数组的大小表示支持的条码格式个数，zxing原本因为支持很多格式，因此这个数组长度比较长。当拿到相机的一帧数据后，需要去检测是否是所有支持格式的某一个格式，每一种格式的检测都需要花费一些时间，因此这个遍历对于我们是不必要的。如果将zxing内部定制成只支持QR Code格式，那么就免去了额外的格式检测。

扫描区域放大到全局

去除项目中的扫描区域，将图像识别区域放大到整张区域，这样增加了二维码的边界情况，不需要特意的对准屏幕的扫描区域。

将相机升级到jetpack的CameraX

谷歌已经在官方提供了对于camera2的整合包，集成在CamreaX，而且CameraX内部有对于图片分析的接口，所以我们在这个接口中会对原来的二维码扫描进行一次转移，然后构建一个线程池专门去处理二维码扫描的分析器。

class CameraXModule(private val view: AutoZoomScanView) {

private var lensFacing: Int = CameraSelector.LENS_FACING_BACK
private var preview: Preview? = null
private var imageAnalyzer: ImageAnalysis? = null
private lateinit var cameraExecutor: ExecutorService
private var camera: Camera? = null
private lateinit var qrCodeAnalyzer: QRCodeAnalyzer
private lateinit var mLifecycleOwner: LifecycleOwner


fun bindWithCameraX(function: (Result) -> Unit, lifecycleOwner: LifecycleOwner) {
    mLifecycleOwner = lifecycleOwner
    val metrics = DisplayMetrics().also { view.display.getRealMetrics(it) }
    Log.d(TAG, "Screen metrics: ${metrics.widthPixels} x ${metrics.heightPixels}")
    val screenAspectRatio = aspectRatio(metrics.widthPixels, metrics.heightPixels)
    Log.i(TAG, "Preview aspect ratio: $screenAspectRatio")
    val cameraSelector = CameraSelector.Builder().requireLensFacing(lensFacing).build()
    val cameraProviderFuture = ProcessCameraProvider.getInstance(view.context)
    cameraProviderFuture.addListener(
        Runnable {
            val cameraProvider: ProcessCameraProvider = cameraProviderFuture.get()
            // Preview
            val width = (view.measuredWidth * 1.5F).toInt()
            val height = (width * screenAspectRatio).toInt()
            preview = Preview.Builder()
                // We request aspect ratio but no resolution
                .setTargetResolution(Size(width, height))
                // Set initial target rotation
                .build()
            preview?.setSurfaceProvider(view.preView.createSurfaceProvider(null))

            cameraExecutor = Executors.newSingleThreadExecutor()
            qrCodeAnalyzer = QRCodeAnalyzer(this) { function(it) }
            // ImageAnalysis
            imageAnalyzer = ImageAnalysis.Builder()
                // We request aspect ratio but no resolution
                .setTargetResolution(Size(width, height))
                // Set initial target rotation, we will have to call this again if rotation changes
                // during the lifecycle of this use case
                .build()
                // The analyzer can then be assigned to the instance
                .also {
                    it.setAnalyzer(cameraExecutor, qrCodeAnalyzer)
                }

            // Must unbind the use-cases before rebinding them
            cameraProvider.unbindAll()

            try {
                // A variable number of use-cases can be passed here -
                // camera provides access to CameraControl & CameraInfo

                camera = cameraProvider.bindToLifecycle(
                    mLifecycleOwner, cameraSelector, preview, imageAnalyzer
                )
                qrCodeAnalyzer.camera = camera
                qrCodeAnalyzer.preview = preview
                setFocus(view.width.toFloat() / 2, view.height.toFloat() / 2)
                // camera?.cameraControl?.startFocusAndMetering(FocusMeteringAction.FLAG_AF)
                // Attach the viewfinder's surface provider to preview use case
            } catch (exc: Exception) {
                Log.e(TAG, "Use case binding failed", exc)
            }

        }, ContextCompat.getMainExecutor(view.context)
    )
}

fun setFocus(x: Float, y: Float) {
    val factory: MeteringPointFactory = SurfaceOrientedMeteringPointFactory(
        view.width.toFloat(), view.height.toFloat()
    )
    //create a point on the center of the view
    val autoFocusPoint = factory.createPoint(x, y)

    camera?.cameraControl?.startFocusAndMetering(
        FocusMeteringAction.Builder(
            autoFocusPoint,
            FocusMeteringAction.FLAG_AF
        ).apply {
            //auto-focus every 1 seconds
            setAutoCancelDuration(1, TimeUnit.SECONDS)
        }.build()
    )
}

private fun aspectRatio(width: Int, height: Int): Double {
    val previewRatio = max(width, height).toDouble() / min(width, height)
    if (abs(previewRatio - RATIO_4_3_VALUE) <= abs(previewRatio - RATIO_16_9_VALUE)) {
        return RATIO_4_3_VALUE
    }
    return RATIO_16_9_VALUE
}

@SuppressLint("RestrictedApi")
fun setZoomRatio(zoomRatio: Float) {
    if (zoomRatio > getMaxZoomRatio()) {
        return
    }
    val future: ListenableFuture<Void>? = camera?.cameraControl?.setZoomRatio(
        zoomRatio
    )
    future?.apply {
        Futures.addCallback(future, object : FutureCallback<Void?> {
            override fun onSuccess(result: Void?) {}
            override fun onFailure(t: Throwable) {}
        }, CameraXExecutors.directExecutor())
    }
}

fun getZoomRatio(): Float {
    return camera?.cameraInfo?.zoomState?.value?.zoomRatio ?: 0F
}

fun getMaxZoomRatio(): Float {
    return camera?.cameraInfo?.zoomState?.value?.maxZoomRatio ?: 0F
}

fun stopCamera() {
    //   camera?.cameraControl?.
}

internal fun resetAnalyzer() {
    qrCodeAnalyzer.resetAnalyzer()
}

companion object {
    private const val TAG = "CameraXImp"
    private const val RATIO_4_3_VALUE = 4.0 / 3.0
    private const val RATIO_16_9_VALUE = 16.0 / 9.0
}
}

上述代码基于的是CameraX内的CameraView，其中的构建的宽高必须基于4:3或者16:9的格式。

自动放大

当二维码很小很远时，自动放大能大大加快检测二维码的速度。QRCodeReader的decode()是二维码检测的主方法，分为两步：

（1）大致判断是否存在二维码；

    val source = PlanarYUVLuminanceSource(data, width, height, 0, 0, width, height, false)
    val binarizer = HybridBinarizer(source)
    val bitmap = BinaryBitmap(binarizer)
    val detectorResult = Detector(bitmap.blackMatrix).detect(map)

   private fun calculateDistance(resultPoint: Array<ResultPoint>): Int {
    val point1X = resultPoint[0].x.toInt()
    val point1Y = resultPoint[0].y.toInt()
    val point2X = resultPoint[1].x.toInt()
    val point2Y = resultPoint[1].y.toInt()
    return sqrt(
        (point1X - point2X.toDouble()).pow(2.0) + (point1Y - point2Y.toDouble()).pow(2.0)
    ).toInt()
}

先要获取到当前区域内是否存在二维码，其次计算二维码的距离。

（2）所以我们需要做的就是先检测该图像区域内是否有一个二维码，同时计算二维码的大小，和图像比例进行一次大小换算，如果发现二维码过小的情况下，自动放大图片区域。

 private fun zoomCamera(points: Array<ResultPoint>, image: BinaryBitmap): Boolean {
    val qrWidth = calculateDistance(points) * 2
    val imageWidth = image.blackMatrix.width.toFloat()
    val zoomInfo = camera?.cameraInfo?.zoomState?.value
    zoomInfo?.apply {
        if (qrWidth < imageWidth / 8) {
            Log.i("BarcodeAnalyzer", "resolved!!! = $qrWidth  imageWidth:${imageWidth}")
            val maxScale = zoomInfo.maxZoomRatio
            val curValue = zoomInfo.zoomRatio
            val gap = maxScale - curValue
            val upgradeRatio = if (gap / 4F * 3 > 3F) 3F else maxScale / 4F * 3
            module.setZoomRatio(curValue + upgradeRatio)
            return true
        }
    }
    return false
}

双击放大

当前二维码扫描中没有调整焦距的功能，所以我们在这次调整中对其进行了一次双击放大的开发。

通过监控双击事件实现对应监听。

 private val gestureDetector =
    GestureDetector(context, object : GestureDetector.SimpleOnGestureListener() {
        override fun onDoubleTap(e: MotionEvent?): Boolean {
            cameraXModule.setZoomRatio(cameraXModule.getZoomRatio() + 1)
            return super.onDoubleTap(e)
        }

        override fun onSingleTapUp(e: MotionEvent?): Boolean {
            e?.apply {
                cameraXModule.setFocus(x, y)
            }
            return super.onSingleTapUp(e)
        }
    })

单击对焦

当前的对焦模式采取的是自动对焦，我们对对焦进行了一次增强，单击制定位置之后会对该区域进行一次对焦。

参考上面代码

简单使用

引入依赖

implementation 'com.github.leifzhang:QrCodeLibrary:0.0.1'

在布局xml中加入AutoZoomScanView

    <com.kronos.camerax.qrcode.AutoZoomScanView
        android:id="@+id/scanView"
        android:layout_width="match_parent"
        android:layout_height="match_parent" />

先申请camera权限并绑定lifecycle

    AndPermission.with(this)
            .runtime()
            .permission(Permission.Group.CAMERA)
            .onGranted { permissions: List<String?>? ->
                scanView.bindWithLifeCycle(this@MainActivity)
            }
            .onDenied { permissions: List<String?>? -> }
            .start()

二维码结果回调，之后重新打开分析逻辑

 scanView.setOnQrResultListener { view: View, s: String ->
            Toast.makeText(
                this@MainActivity, s,
                Toast.LENGTH_LONG
            ).show()
            scanView.reStart()
        }

本文参与腾讯云自媒体同步曝光计划，分享自作者个人站点/博客。

原始发表：2024-01-27，如有侵权请联系 cloudcommunity@tencent.com 删除

二维码