Skip to main content

ImageQualityInterpretation

Parent: Options

When document images don't meet our image requirements fully, they can end up falsely rejected. This happens because some checks require high quality images, or because it's not possible to differentiate image quality issues from genuine fraud.

For example, if the document number is obscured, a few data checks will fail. Verify can't know what it says under the obscurity and whether the end user is trying to hide something.

Similarly, there may be image quality issues that don't cause anything to be misread, but they could still be used to obscure fraud like image editing, or physical tampering. Depending on your business and use case, you might care more about some of these situations than others.

For example, if you're extremely sensitive to fraud, and have a High Assurance use case, you might not want to allow users to pass with suboptimal image quality, even if all fraud checks seem to be passing.

On the other hand, if you're optimizing for maximum user conversion, you might not want to immediately accuse users of fraud if fraud checks might have failed because of suboptimal image quality.

That's where the ImageQualityInterpretation option comes in. It allows you to specify how Verify should interpret the image quality issues, especially when they occur in conjuction with failing fraud checks.

info

Some checks are definitive and aren't affected by image quality issues. For example, if the barcode can be read, we can run barcode data checks on it without considering the quality of the image at all. Those definitive checks are always trusted, and the ImageQualityInterpretation option doesn't affect them.

Possible Values

ValueDescription
IgnoreVerify will always make a fraud verdict (for supported document types)
ConservativeVerify won't make Pass nor Fail verdicts if image quality issues are detected
HighAssuranceVerify will Fail documents even if they have image quality issues, but won't Pass them if there are image quality issues
HighConversionVerify won't make Fail verdicts if image quality issues are detected
VeryHighConversionVerify will tolerate even more issues instead of making explicit Fail verdicts (you should have a very good reason to use this)

Verdicts won't ever map into the opposite (e.g. Fail won't become Pass or vice versa). Instead, if Verify can't make a fraud verdict, the overall fraud check will be NotPerformed.

The RecommendedOutcome feature takes this into account and will get mapped in the following way (provided the document type is supported):

OverallFraudCheckRecommendedOutcome
PassAccept
FailReject
NotPerformedRetry

Conservative is the default value.