Extract document images

BlinkID can return images alongside extraction results. This is useful when you need to pass the captured document images to a downstream service—such as the BlinkID Verify API—or store them for audit purposes.

Types of images

BlinkID can return two distinct types of images:

• Input images (raw camera frames): the original, unprocessed frames captured by the camera, including the background around the document.
• Document images (cropped): perspective-corrected images tightly framed on the document itself.

If you are passing images to Verify, use raw input images.

Enable image return

Web (TS)

Set returnInputImages: true in your scanning settings:

const settings: ScanningSettings = {
  returnInputImages: true,
};

To also receive cropped document or face images, configure croppedImageSettings:

const settings: ScanningSettings = {
  returnInputImages: true,
  croppedImageSettings: {
    returnDocumentImage: true,
    returnFaceImage: true,
  },
};

Android (Kotlin)

Set returnInputImages = true inside ScanningSettings:

val sessionSettings = BlinkIdSessionSettings(
    scanningSettings = ScanningSettings(
        returnInputImages = true
    )
)

Pass sessionSettings when launching the scanning screen:

BlinkIdCameraScanningScreen(
    blinkIdSdk = sdk,
    sessionSettings = sessionSettings,
    onScanningSuccess = { result -> /* handle result */ },
    onScanningCancelled = { /* handle cancellation */ }
)

iOS (Swift)

Set returnInputImages: true inside ScanningSettings:

let analyzer = try await BlinkIDAnalyzer(
    sdk: sdkInstance,
    blinkIdSessionSettings: BlinkIDSessionSettings(
        scanningSettings: ScanningSettings(
            returnInputImages: true
        )
    ),
    eventStream: BlinkIDEventStream()
)

Access images on the result

Web (TS)

The result contains inputImageFront, inputImageBack, and inputImageBarcode fields with the raw ImageData frames.

Encode to PNG before sending to an API:

const canvas = document.createElement("canvas");
canvas.width = result.inputImageFront.width;
canvas.height = result.inputImageFront.height;
canvas.getContext("2d")!.putImageData(result.inputImageFront, 0, 0);
const base64 = canvas.toDataURL("image/png").split(",")[1];

Android (Kotlin)

Call inputImage(scanningSide) on the result to get the raw frame for each side:

val frontBitmap = scanningResult.inputImage(ScanningSide.FIRST)?.bitmap
val backBitmap = scanningResult.inputImage(ScanningSide.SECOND)?.bitmap

fun encodeToPng(bitmap: Bitmap): String {
    val stream = ByteArrayOutputStream()
    bitmap.compress(Bitmap.CompressFormat.PNG, 100, stream)
    return Base64.encodeToString(stream.toByteArray(), Base64.DEFAULT)
}

iOS (Swift)

Call getInputImage(scanningSide:) on the result to get the raw frame for each side:

if let front = scanningResult.getInputImage(scanningSide: .first),
   let pngData = front.uiImage?.pngData() {
    let base64 = pngData.base64EncodedString()
    // pass base64 to your API
}

if let back = scanningResult.getInputImage(scanningSide: .second),
   let pngData = back.uiImage?.pngData() {
    let base64 = pngData.base64EncodedString()
}

Use lossless compression

Do not use lossy compression (such as JPEG) on images you intend to pass to a verification backend. Lossy compression degrades image quality in ways that can affect analysis. Encode to PNG instead, as shown in the examples above.

Do not post-process images

Pass the images as-is from the SDK result. Do not crop, resize, rotate, or otherwise modify them after capture. Post-processing can break assumptions that downstream services make about image integrity.