MRZ incorrect recognition

Issue

A document with a Machine-Readable Zone (MRZ) cannot be recognized, or the MRZ is recognized improperly. This issue may occur when processing documents through the Web API or other platforms where infrared (IR) light is unavailable.

Solution

To improve MRZ recognition, follow these steps based on the method of document processing:

1. Physical Readers: If you are using physical readers like 1120 or 7310 devices, the SDK reads the MRZ using infrared (IR) light. This ensures that the MRZ is visible while other parts of the document become invisible, minimizing interference.
2. Web API Processing: When processing documents through the Web API, the SDK uses white light, which can lead to interference from other elements in the MRZ area. To address this, use the following parameter to enhance MRZ recognition:

   {"mrzDetectMode": 1}

   This mode adjusts the binarization of the image to improve MRZ detection in scenarios where IR light is unavailable, such as web-based document processing.

   You can test this feature here, by entering the string {"mrzDetectMode": 1} in the Custom Parameters field, as shown in the screenshot below:

   

3. Robust Detection Mode: If the image quality is extremely poor or the background is excessively complex, use the following parameter for even more robust detection:

   {"mrzDetectMode": 2}

   This mode is designed to handle challenging scenarios and ensure accurate MRZ recognition.

Process Sequence

The mrzDetectMode parameter operates according to the following logic:

1. If the document has an MRZ and it was successfully read in standard mode, the mrzDetectMode will not be applied.
2. If the document has an MRZ but it was not read in standard mode, the mrzDetectMode will be applied to improve recognition.
3. If the document does not contain an MRZ, the mrzDetectMode will not be applied.