Apple’s Vision framework contains computer vision related functionality and with iOS 13 it can detect text on images as well. Moreover, Apple added a new framework VisionKit what makes it easy to integrate document scanning functionality. For demonstrating the usage of it, let’s build a simple UI what can present the scanner and display scanned text.
Scanning text with Vision
VisionKit has VNDocumentCameraViewController and when presented, it allows scanning documents and cropping scanned documents. It uses delegate for publishing scanned documents via an instance of VNDocumentCameraScan. This object contains all the taken images (documents). Next, we can use VNImageRequestHandler in Vision for detecting text on those images.
final class TextRecognizer {
let cameraScan: VNDocumentCameraScan
init(cameraScan: VNDocumentCameraScan) {
self.cameraScan = cameraScan
}
private let queue = DispatchQueue(label: "com.augmentedcode.scan", qos: .default, attributes: [], autoreleaseFrequency: .workItem)
func recognizeText(withCompletionHandler completionHandler: @escaping ([String]) -> Void) {
queue.async {
let images = (0..<self.cameraScan.pageCount).compactMap({ self.cameraScan.imageOfPage(at: $0).cgImage })
let imagesAndRequests = images.map({ (image: $0, request: VNRecognizeTextRequest()) })
let textPerPage = imagesAndRequests.map { image, request -> String in
let handler = VNImageRequestHandler(cgImage: image, options: [:])
do {
try handler.perform([request])
guard let observations = request.results as? [VNRecognizedTextObservation] else { return "" }
return observations.compactMap({ $0.topCandidates(1).first?.string }).joined(separator: "\n")
}
catch {
print(error)
return ""
}
}
DispatchQueue.main.async {
completionHandler(textPerPage)
}
}
}
}
Presenting document scanner with SwiftUI
As VNDocumentCameraViewController is UIKit view controller we can’t directly present it in SwiftUI. For making this work, we’ll need to use a separate value type conforming to UIViewControllerRepresentable protocol. UIViewControllerRepresentable is the glue between SwiftUI and UIKit and enables us to present UIKit views. This protocol requires us to define the class of the view controller and then implementing makeUIViewController(context:) and updateUIViewController(_:context:). In addition, we’ll also create coordinator what is going to be VNDocumentCameraViewController’s delegate. SwiftUI uses UIViewRepresentableContext for holding onto the coordinator and managing the view controller updates behind the scenes. Our case is pretty simple, we just use completion handler for passing back scanned text or nil when it was closed or error occurred. No need to update the view controller itself, only to pass data from it back to SwiftUI.
struct ScannerView: UIViewControllerRepresentable {
private let completionHandler: ([String]?) -> Void
init(completion: @escaping ([String]?) -> Void) {
self.completionHandler = completion
}
typealias UIViewControllerType = VNDocumentCameraViewController
func makeUIViewController(context: UIViewControllerRepresentableContext<ScannerView>) -> VNDocumentCameraViewController {
let viewController = VNDocumentCameraViewController()
viewController.delegate = context.coordinator
return viewController
}
func updateUIViewController(_ uiViewController: VNDocumentCameraViewController, context: UIViewControllerRepresentableContext<ScannerView>) {}
func makeCoordinator() -> Coordinator {
return Coordinator(completion: completionHandler)
}
final class Coordinator: NSObject, VNDocumentCameraViewControllerDelegate {
private let completionHandler: ([String]?) -> Void
init(completion: @escaping ([String]?) -> Void) {
self.completionHandler = completion
}
func documentCameraViewController(_ controller: VNDocumentCameraViewController, didFinishWith scan: VNDocumentCameraScan) {
print("Document camera view controller did finish with ", scan)
let recognizer = TextRecognizer(cameraScan: scan)
recognizer.recognizeText(withCompletionHandler: completionHandler)
}
func documentCameraViewControllerDidCancel(_ controller: VNDocumentCameraViewController) {
completionHandler(nil)
}
func documentCameraViewController(_ controller: VNDocumentCameraViewController, didFailWithError error: Error) {
print("Document camera view controller did finish with error ", error)
completionHandler(nil)
}
}
}
ContentView is the main SwiftUI view presenting the content for our very simple UI with static text, button and scanned text. When pressing on the button, we’ll set isShowingScannerSheet property to true. As it is @State property, then this change triggers SwiftUI update and sheet modifier will take care of presenting ScannerView with VNDocumentCameraViewController. When view controller finishes, completion handler is called and we will update the text property and set isShowingScannerSheet to false which triggers tearing down the modal during the next update.
struct ContentView: View {
private let buttonInsets = EdgeInsets(top: 8, leading: 16, bottom: 8, trailing: 16)
var body: some View {
VStack(spacing: 32) {
Text("Vision Kit Example")
Button(action: openCamera) {
Text("Scan").foregroundColor(.white)
}.padding(buttonInsets)
.background(Color.blue)
.cornerRadius(3.0)
Text(text).lineLimit(nil)
}.sheet(isPresented: self.$isShowingScannerSheet) { self.makeScannerView() }
}
@State private var isShowingScannerSheet = false
@State private var text: String = ""
private func openCamera() {
isShowingScannerSheet = true
}
private func makeScannerView() -> ScannerView {
ScannerView(completion: { textPerPage in
if let text = textPerPage?.joined(separator: "\n").trimmingCharacters(in: .whitespacesAndNewlines) {
self.text = text
}
self.isShowingScannerSheet = false
})
}
}
Summary
With the new addition of VisionKit and text recognition APIs, it is extremely easy to add support of scanning text using camera.
If this was helpful, please let me know on Mastodon@toomasvahter or Twitter @toomasvahter. Feel free to subscribe to RSS feed. Thank you for reading.
Example
VisionKitExample (Xcode 11)
Augmented Code 