Bank Cheque OCR API Integration: Fields and Validation

A cheque OCR integration converts an image into fields an application can inspect. Useful fields include the cheque number, date, payee and amounts in words and numerals. The application also needs to know whether extraction ran, which required values are missing and which configured checks reported an issue.

Azimut’s scanner integration can include an OCR result with separate content and validation objects. The source-backed response example shows an amount mismatch and the review decision it should trigger. It is a data example, not a hosted OCR service address. Obtain the runtime-specific endpoint, authentication requirements and full contract from the integration team before writing a client.

Agree the capture and result contract

Confirm the scanner or image source, operating system, image sides and formats, and the required cheque fields. Optical text, MICR data and validation results have different meanings. Preserve those distinctions when mapping the response into the application’s item record.

A successful scan may have no OCR result when that processing step is unavailable or disabled. Likewise, a false error flag alone is insufficient evidence that a particular check ran. Agree how the deployed contract represents absent or incomplete checks and route uncertain items to review. Field confidence is available only where the configured processing integration supplies it; do not infer it from a device quality score.

Evaluate before connecting posting

Use labelled cheque images representative of the intended handwriting, stock and capture conditions. Measure results by field and include incorrect automatic accepts, review rate and capture failures. Keep the acceptance set separate from configuration samples. Batch support, throughput and processing time should be evaluated for the selected deployment rather than assumed from a single-item example.

Once an item is approved, the application can pass it to the agreed banking, clearing or ERP interface. Retain its original identity and downstream acknowledgement so a timeout can be reconciled before retrying. See the capture-to-posting evaluation for the failure cases to demonstrate.

Agree data location, secure transfer, access and retention for the complete processing path. Request an integration walkthrough with the capture configuration and target interface; arrange secure sample transfer before sending real cheque images.