Building Desktop Passport Scanner with Qt and USB Camera

If you search for passport scanner software or MRZ reader software, you will find many of them are only available for mobile devices. For police officers, scanning passports by mobile devices is convenient when they are patrolling. However, for customs and immigration officers, they usually use desktop system and professional passport scanner or reader, which cost a lot, to check passengers’ passport information. Dynamsoft’s OCR SDK is available for both mobile and desktop scenarios. In this article, I will demonstrate an economic way that uses a cheap USB web camera (less than $20), Qt, and Dynamsoft MRZ SDK to build a desktop passport scanner application for Windows and Linux.

Prerequisites

  • Qt 5.12.11
    • Windows
    • Linux

        sudo apt-get install qt5-default
      

The Skeleton of Qt C++ Project for Desktop Passport Scanner

Before getting started, let’s get the codebase of the barcode scanning application that I implemented recently.

git clone https://github.com/yushulx/Qt-desktop-barcode-reader.git

The codebase has implemented the file loading and camera video streaming functions. What I need to do is to replace barcode recognition SDK with MRZ recognition SDK. In addition, the project needs to import extra character models (trained by Caffe) for OCR and a template file for providing MRZ recognition parameters.

Character model files

NumberUppercase.caffemodel
NumberUppercase.prototxt
NumberUppercase.txt
NumberUppercase_Assist_1lIJ.caffemodel
NumberUppercase_Assist_1lIJ.prototxt
NumberUppercase_Assist_1lIJ.txt
NumberUppercase_Assist_8B.caffemodel
NumberUppercase_Assist_8B.prototxt
NumberUppercase_Assist_8B.txt
NumberUppercase_Assist_8BHR.caffemodel
NumberUppercase_Assist_8BHR.prototxt
NumberUppercase_Assist_8BHR.txt
NumberUppercase_Assist_number.caffemodel
NumberUppercase_Assist_number.prototxt
NumberUppercase_Assist_number.txt
NumberUppercase_Assist_O0DQ.caffemodel
NumberUppercase_Assist_O0DQ.prototxt
NumberUppercase_Assist_O0DQ.txt
NumberUppercase_Assist_upcase.caffemodel
NumberUppercase_Assist_upcase.prototxt
NumberUppercase_Assist_upcase.txt

Template file

{
   "CharacterModelArray" : [
    {
      "DirectoryPath": "CharacterModel",
      "FilterFilePath": "",
      "Name": "NumberUppercase"
    }
   ],
   "LabelRecognizerParameterArray" : [
      {
         "BinarizationModes" : [
            {
               "BlockSizeX" : 0,
               "BlockSizeY" : 0,
               "EnableFillBinaryVacancy" : 1,
               "LibraryFileName" : "",
               "LibraryParameters" : "",
               "Mode" : "BM_LOCAL_BLOCK",
               "ThreshValueCoefficient" : 15
            }
         ],
         "CharacterModelName" : "NumberUppercase",
         "LetterHeightRange" : [ 5, 1000, 1 ],
		 "LineStringLengthRange" : [44, 44],
		 "MaxLineCharacterSpacing" : 130,
		 "LineStringRegExPattern" : "(P[OM<][A-Z]{3}([A-Z<]{0,35}[A-Z]{1,3}[(<<)][A-Z]{1,3}[A-Z<]{0,35}<{0,35}){(39)}){(44)}|([A-Z0-9<]{9}[0-9][A-Z]{3}[0-9]{2}[(01-12)][(01-31)][0-9][MF][0-9]{2}[(01-12)][(01-31)][0-9][A-Z0-9<]{14}[0-9<][0-9]){(44)}",
         "MaxThreadCount" : 4,
         "Name" : "locr",
		 "TextureDetectionModes" :[
			{
				"Mode" : "TDM_GENERAL_WIDTH_CONCENTRATION",
				"Sensitivity" : 8
			}
		 ],
         "ReferenceRegionNameArray" : [ "DRRegion" ]
      }
   ],
   "LineSpecificationArray" : [
    {
		"Name":"L0",
		"LineNumber":"",
		"BinarizationModes" : [
            {
               "BlockSizeX" : 30,
               "BlockSizeY" : 30,
               "Mode" : "BM_LOCAL_BLOCK"
            }
         ]
    }
	],
   "ReferenceRegionArray" : [
      {
         "Localization" : {
            "FirstPoint" : [ 0, 0 ],
            "SecondPoint" : [ 100, 0 ],
            "ThirdPoint" : [ 100, 100 ],
            "FourthPoint" : [ 0, 100 ],
            "MeasuredByPercentage" : 1,
            "SourceType" : "LST_MANUAL_SPECIFICATION"
         },
         "Name" : "DRRegion",
         "TextAreaNameArray" : [ "DTArea" ]
      }
   ],
   "TextAreaArray" : [
      {
         "LineSpecificationNameArray" : ["L0"],
         "Name" : "DTArea",
		 "FirstPoint" : [ 0, 0 ],
         "SecondPoint" : [ 100, 0 ],
         "ThirdPoint" : [ 100, 100 ],
		 "FourthPoint" : [ 0, 100 ]
      }
   ]
}

Desktop Passport Scanner for Windows and Linux

Since we have already got the codebase, it won’t take too much time to get the application work.

Library Linking

We extract the library files from the downloaded archive and put them into the corresponding folders:

  • Windows

    Copy DynamsoftLabelRecognizerx64.lib to platform/windows/lib.

    Copy DynamicPdfx64.dll, DynamsoftLabelRecognizerx64.dll, DynamsoftLicenseClientx64.dll and vcomp140.dll to platform/windows/bin.

  • Linux

    Copy libDynamicPdf.so, libDynamsoftLabelRecognizer.so, and libDynamsoftLicenseClient.so to platform/linux.

After that, we update the CMakeLists.txt file to link libraries and copy model and template files:

if (CMAKE_HOST_WIN32)
    target_link_libraries(${PROJECT_NAME} PRIVATE Qt5::Widgets Qt5::MultimediaWidgets "DynamsoftLabelRecognizerx64")    
elseif(CMAKE_HOST_UNIX)
    target_link_libraries(${PROJECT_NAME} PRIVATE Qt5::Widgets Qt5::MultimediaWidgets "DynamsoftLabelRecognizer")
endif()

# Copy DLLs
if(CMAKE_HOST_WIN32)
    add_custom_command(TARGET ${PROJECT_NAME} POST_BUILD 
        COMMAND ${CMAKE_COMMAND} -E copy_directory
        "${PROJECT_SOURCE_DIR}/platform/windows/bin/"      
        $<TARGET_FILE_DIR:${PROJECT_NAME}>)
endif()

# Copy template
add_custom_command(TARGET ${PROJECT_NAME} POST_BUILD 
        COMMAND ${CMAKE_COMMAND} -E copy_directory
        "${PROJECT_SOURCE_DIR}/template/"      
        $<TARGET_FILE_DIR:${PROJECT_NAME}>)

# Copy model files
add_custom_command(TARGET ${PROJECT_NAME} POST_BUILD 
        COMMAND ${CMAKE_COMMAND} -E copy_directory
        "${PROJECT_SOURCE_DIR}/CharacterModel"      
        $<TARGET_FILE_DIR:${PROJECT_NAME}>/CharacterModel)

Steps to Modify the Code for MRZ Recognition

Next, we import DynamsoftLabelRecognizer.h and DynamsoftCore.h to mainwindow.h:

#include "DynamsoftLabelRecognizer.h"
#include "DynamsoftCore.h"

In mainwindow.cpp, we search for the line of invoking DBR_DecodeFile() method and then replace the line with:

int errorCode = DBR_DecodeFile(reader, fileName.toStdString().c_str(), "");

Add the following code block to get passport information:

DLR_Result **results = handler->results;
for (int ri = 0; ri < handler->resultsCount; ++ri)
{
    DLR_Result* result = handler->results[ri];
    int lCount = result->lineResultsCount;
    for (int li = 0; li < lCount; ++li)
    {
        DM_Point *points = result->lineResults[li]->location.points;
        int x1 = points[0].x, y1 = points[0].y;
        int x2 = points[1].x, y2 = points[1].y;
        int x3 = points[2].x, y3 = points[2].y;
        int x4 = points[3].x, y4 = points[3].y;
    }

    if (lCount < 2)
    {
        continue;
    }
    string line1 = result->lineResults[0]->text;
    string line2 = result->lineResults[1]->text;
    if (line1.length() != 44 || line2.length() != 44)
    {
        continue;
    }
    if (line1[0] != 'P')
        continue;
    else {
        // Type
        string tmp = "Type: ";
        tmp.insert(tmp.length(), 1, line1[0]);
        out += QString::fromStdString(tmp) + "\n";

        // Issuing country
        tmp = "Issuing country: "; line1.substr(2, 5);
        tmp += line1.substr(2, 3);		
        out += QString::fromStdString(tmp) + "\n";

        // Surname
        int index = 5;
        tmp = "Surname: ";
        for (; index < 44; index++)
        {
            if (line1[index] != '<')
            {
                tmp.insert(tmp.length(), 1, line1[index]);
            }
            else 
            {
                break;
            }
        }
        out += QString::fromStdString(tmp) + "\n";

        // Given names
        tmp = "Given Names: ";
        index += 2;
        for (; index < 44; index++)
        {
            if (line1[index] != '<')
            {
                tmp.insert(tmp.length(), 1, line1[index]);
            }
            else 
            {
                tmp.insert(tmp.length(), 1, ' ');
            }
        }
        out += QString::fromStdString(tmp) + "\n";

        // Passport number
        tmp = "Passport number: ";
        index = 0;
        for (; index < 9; index++)
        {
            if (line2[index] != '<')
            {
                tmp.insert(tmp.length(), 1, line2[index]);
            }
            else 
            {
                break;
            }
        }
        out += QString::fromStdString(tmp) + "\n";

        // Nationality
        tmp = "Nationality: ";
        tmp += line2.substr(10, 3);
        out += QString::fromStdString(tmp) + "\n";

        // Date of birth
        tmp = line2.substr(13, 6);
        tmp.insert(2, "/");
        tmp.insert(5, "/");
        tmp = "Date of birth (YYMMDD): " + tmp;
        out += QString::fromStdString(tmp) + "\n";

        // Sex
        tmp = "Sex: ";
        tmp.insert(tmp.length(), 1, line2[20]);
        out += QString::fromStdString(tmp) + "\n";

        // Expiration date of passport
        tmp = line2.substr(21, 6);
        tmp.insert(2, "/");
        tmp.insert(5, "/");
        tmp = "Expiration date of passport (YYMMDD): " + tmp;
        out += QString::fromStdString(tmp) + "\n";

        // Personal number
        if (line2[28] != '<')
        {
            tmp = "Personal number: ";
            for (index = 28; index < 42; index++)
            {
                if (line2[index] != '<')
                {
                    tmp.insert(tmp.length(), 1, line2[index]);
                }
                else 
                {
                    break;
                }
            }
            out += QString::fromStdString(tmp) + "\n";
        }
    }
}
DLR_FreeResults(&handler);

So far, the static image recognition is completed. In the following, we will implement real-time passport scanning by camera video stream.

To store MRZ recognition information and share it between threads, we create a new class MRZInfo:

#ifndef MRZINFO_H
#define MRZINFO_H

#include <QString>

class MRZInfo
{
public:
    MRZInfo() = default;

    ~MRZInfo(){};

    bool isNull();

public:
    QString text;
    int x1, y1, x2, y2, x3, y3, x4, y4, xx1, yy1, xx2, yy2, xx3, yy3, xx4, yy4;
};

#endif // MRZINFO_H

Open work.h to add a new slot function detectMRZ(), which works in a worker thread for recognizing MRZ:

void Work::detectMRZ()
{
    while (m_bIsRunning)
    {
        QImage image;
        m_mutex.lock();
        // wait for QList
        if (queue.isEmpty())
        {
            m_listIsEmpty.wait(&m_mutex);
        }

        if (!queue.isEmpty())
        {
            image = queue.takeFirst();
        }
        m_mutex.unlock();

        if (!image.isNull())
        {
            // Detect MRZ
        }
    }
}

To recognize MRZ and extract passport information, we firstly convert QImage to ImageData and then call DLR_RecognizeByBuffer() method:

// Convert QImage to ImageData
ImageData data;
data.bytes = (unsigned char *)image.bits();
data.width = image.width();
data.height = image.height();
data.stride = image.bytesPerLine();
data.format = IPF_ARGB_8888;
data.bytesLength = image.byteCount();

QDateTime start = QDateTime::currentDateTime();
int errorCode = DLR_RecognizeByBuffer(recognizer, &data, "locr");
QDateTime end = QDateTime::currentDateTime();
DLR_ResultArray *handler = NULL;
DLR_GetAllResults(recognizer, &handler);
std::vector<MRZInfo> all;
QString out = "Elapsed time: " + QString::number(start.msecsTo(end)) + "ms\n\n";
DLR_Result **results = handler->results;
for (int ri = 0; ri < handler->resultsCount; ++ri)
{
    DLR_Result* result = handler->results[ri];
    int lCount = result->lineResultsCount;
    if (lCount < 2)
    {
        continue;
    }

    DLR_LineResult *l1 = result->lineResults[0];
    DLR_LineResult *l2 = result->lineResults[1];

    string line1 = l1->text;
    string line2 = l2->text;
    if (line1.length() != 44 || line2.length() != 44)
    {
        continue;
    }
    if (line1[0] != 'P')
        continue;

    MRZInfo info;
    
    DM_Point *points = l1->location.points;
    int x1 = points[0].x, y1 = points[0].y;
    int x2 = points[1].x, y2 = points[1].y;
    int x3 = points[2].x, y3 = points[2].y;
    int x4 = points[3].x, y4 = points[3].y;
    DM_Point *points2 = l2->location.points;
    int xx1 = points2[0].x, yy1 = points2[0].y;
    int xx2 = points2[1].x, yy2 = points2[1].y;
    int xx3 = points2[2].x, yy3 = points2[2].y;
    int xx4 = points2[3].x, yy4 = points2[3].y;

    // Type
    string tmp = "Type: ";
    tmp.insert(tmp.length(), 1, line1[0]);
    out += QString::fromStdString(tmp) + "\n";

    // Issuing country
    tmp = "Issuing country: "; line1.substr(2, 5);
    tmp += line1.substr(2, 3);		
    out += QString::fromStdString(tmp) + "\n";

    // Surname
    int index = 5;
    tmp = "Surname: ";
    for (; index < 44; index++)
    {
        if (line1[index] != '<')
        {
            tmp.insert(tmp.length(), 1, line1[index]);
        }
        else 
        {
            break;
        }
    }
    out += QString::fromStdString(tmp) + "\n";

    // Given names
    tmp = "Given Names: ";
    index += 2;
    for (; index < 44; index++)
    {
        if (line1[index] != '<')
        {
            tmp.insert(tmp.length(), 1, line1[index]);
        }
        else 
        {
            tmp.insert(tmp.length(), 1, ' ');
        }
    }
    out += QString::fromStdString(tmp) + "\n";

    // Passport number
    tmp = "Passport number: ";
    index = 0;
    for (; index < 9; index++)
    {
        if (line2[index] != '<')
        {
            tmp.insert(tmp.length(), 1, line2[index]);
        }
        else 
        {
            break;
        }
    }
    out += QString::fromStdString(tmp) + "\n";

    // Nationality
    tmp = "Nationality: ";
    tmp += line2.substr(10, 3);
    out += QString::fromStdString(tmp) + "\n";

    // Date of birth
    tmp = line2.substr(13, 6);
    tmp.insert(2, "/");
    tmp.insert(5, "/");
    tmp = "Date of birth (YYMMDD): " + tmp;
    out += QString::fromStdString(tmp) + "\n";

    // Sex
    tmp = "Sex: ";
    tmp.insert(tmp.length(), 1, line2[20]);
    out += QString::fromStdString(tmp) + "\n";

    // Expiration date of passport
    tmp = line2.substr(21, 6);
    tmp.insert(2, "/");
    tmp.insert(5, "/");
    tmp = "Expiration date of passport (YYMMDD): " + tmp;
    out += QString::fromStdString(tmp) + "\n";

    // Personal number
    if (line2[28] != '<')
    {
        tmp = "Personal number: ";
        for (index = 28; index < 42; index++)
        {
            if (line2[index] != '<')
            {
                tmp.insert(tmp.length(), 1, line2[index]);
            }
            else 
            {
                break;
            }
        }
        out += QString::fromStdString(tmp) + "\n";
    }

    info.text = out;
    info.x1 = x1;
    info.y1 = y1;
    info.x2 = x2;
    info.y2 = y2;
    info.x3 = x3;
    info.y3 = y3;
    info.x4 = x4;
    info.y4 = y4;
    info.xx1 = xx1;
    info.yy1 = yy1;
    info.xx2 = xx2;
    info.yy2 = yy2;
    info.xx3 = xx3;
    info.yy3 = yy3;
    info.xx4 = xx4;
    info.yy4 = yy4;

    all.push_back(info);
}

DLR_FreeResults(&handler);
surface->appendResult(all);

How to Build the Qt CMake Project

The CMake build commands are a bit different between Windows and Linux:

# Windows
mkdir build
cd build
cmake -G "MinGW Makefiles" ..
cmake --build .
MRZRecognizer.exe

# Linux
mkdir build
cd build
cmake ..
cmake --build .
./MRZRecognizer

Running Passport Scanner

When running the program, you need to enter a valid license key:

mrz SDK license

Then you can scan passport information from static images or webcam.

passport scanner for Windows and Linux

Source Code

https://github.com/yushulx/passport-scanner

Search Blog Posts