Question

我正在解决扫描面部前置摄像头输入，检测面部并将其作为 UIImage 对象进行扫描的任务。我正在使用 AVFoundation 来扫描和检测面孔。

像这样：

let input = try AVCaptureDeviceInput(device: captureDevice)
captureSession = AVCaptureSession()
captureSession!.addInput(input)

output = AVCaptureMetadataOutput()
captureSession?.addOutput(output)

output.setMetadataObjectsDelegate(self, queue: dispatch_get_main_queue())
output.metadataObjectTypes = [AVMetadataObjectTypeFace]


videoPreviewLayer = AVCaptureVideoPreviewLayer(session: captureSession)
videoPreviewLayer?.videoGravity = AVLayerVideoGravityResizeAspectFill
videoPreviewLayer?.frame = view.layer.bounds
view.layer.addSublayer(videoPreviewLayer!)

captureSession?.startRunning()

在委托方法中 didOutputMetadataObjects 我正在面对 AVMetadataFaceObject 并使用红框高亮显示它：

let metadataObj = metadataObjects[0] as! AVMetadataFaceObject
let faceObject = videoPreviewLayer?.transformedMetadataObjectForMetadataObject(metadataObj)
faceFrame?.frame = faceObject!.bounds

问题是：如何以 UIImages 来获得面孔？

我试图跳过' didOutputSampleBuffer '，但根本没有调用它：c

Answer 1

我使用didOutputSampleBuffer和Objective-C做了同样的事情。它看起来像：

- (void)captureOutput:(AVCaptureOutput *)captureOutput didOutputSampleBuffer:(CMSampleBufferRef)sampleBuffer fromConnection:(AVCaptureConnection *)connection {
  CVPixelBufferRef pixelBuffer = CMSampleBufferGetImageBuffer(sampleBuffer);
  CFDictionaryRef attachments = CMCopyDictionaryOfAttachments(kCFAllocatorDefault, sampleBuffer, kCMAttachmentMode_ShouldPropagate);
  CIImage *ciImage = [[CIImage alloc] initWithCVPixelBuffer:pixelBuffer options:(__bridge NSDictionary *)attachments];
  if (attachments)
    CFRelease(attachments);

  NSNumber *orientation = (__bridge NSNumber *)(CMGetAttachment(imageDataSampleBuffer, kCGImagePropertyOrientation, NULL));
  NSArray *features = [[CIDetector detectorOfType:CIDetectorTypeFace context:nil options:@{ CIDetectorAccuracy: CIDetectorAccuracyHigh }] featuresInImage:ciImage options:@{ CIDetectorImageOrientation: orientation }];
  if (features.count == 1) {
    CIFaceFeature *faceFeature = [features firstObject];
    CGRect faceRect = faceFeature.bounds;

    CGImageRef tempImage = [[CIContext contextWithOptions:nil] createCGImage:ciImage fromRect:ciImage.extent];
    UIImage *image = [UIImage imageWithCGImage:tempImage scale:1.0 orientation:orientation.intValue];
    UIImage *face = [image extractFace:faceRect];
  }  
}

其中extractFace是UIImage的扩展：

- (UIImage *)extractFace:(CGRect)rect {
  rect = CGRectMake(rect.origin.x * self.scale,
                  rect.origin.y * self.scale,
                  rect.size.width * self.scale,
                  rect.size.height * self.scale);
  CGImageRef imageRef = CGImageCreateWithImageInRect(self.CGImage, rect);
  UIImage *result = [UIImage imageWithCGImage:imageRef scale:self.scale orientation:self.imageOrientation];
  CGImageRelease(imageRef);
  return result;
}

制作视频输出：

AVCaptureVideoDataOutput *videoOutput = [[AVCaptureVideoDataOutput alloc] init];
videoOutput.videoSettings = @{ (id)kCVPixelBufferPixelFormatTypeKey: [NSNumber numberWithInt:kCMPixelFormat_32BGRA] };
videoOutput.alwaysDiscardsLateVideoFrames = YES;
self.videoOutputQueue = dispatch_queue_create("OutputQueue", DISPATCH_QUEUE_SERIAL);
[videoOutput setSampleBufferDelegate:self queue:self.videoOutputQueue];
[self.session addOutput:videoOutput];

Answer 2

- (UIImage *) screenshot {

CGSize size = CGSizeMake(faceFrame.frame.size.width, faceFrame.frame.size.height);

UIGraphicsBeginImageContextWithOptions(size, NO, [UIScreen     mainScreen].scale);

CGRect rec = CGRectMake(faceFrame.frame.origin.x, faceFrame.frame.orogin.y, faceFrame.frame.size.width, faceFrame.frame.size.height);
[_viewController.view drawViewHierarchyInRect:rec   afterScreenUpdates:YES];

UIImage *image = UIGraphicsGetImageFromCurrentImageContext();
UIGraphicsEndImageContext();
return image;
}

从上面获取一些提示

  let contextImage: UIImage = <<screenshot>>!
let cropRect: CGRect = CGRectMake(x, y, width, height)
let imageRef: CGImageRef =     CGImageCreateWithImageInRect(contextImage.CGImage, cropRect)
let image: UIImage = UIImage(CGImage: imageRef, scale:   originalImage.scale, orientation: originalImage.imageOrientation)!

Answer 3

我建议使用UIImagePickerController类来实现自定义相机，以便为人脸检测选择多个图像。请检查Apple的示例代码PhotoPicker。

向某些人使用UIImagePickerController将相机作为sourceType启动。并处理其委托imagePickerController:didFinishPickingMediaWithInfo:以捕获图像+如果有帮助，您还可以检查takePicture函数

（howto）从输入到前置摄像头获取面部作为UIImage

3 个答案: