milliyang/directshow_camera

A window based camera library (Directshow) in c++.

Directshow Camera

A window-based camera library in c++.

Features

• Work with OpenCV::Mat
• Support all camera settings
• Support Exposure Fusion
• A looper to handle the camera cycle
• Internal fake camera (Stub) for testing

Table of contents

1. Usage
2. Looper
3. Exposure Fusion
4. Using camera without OpenCV
5. Requirements
6. Installation
7. Deploy example codes
8. License

Usage

#include <uvc_camera.h>

using namespace DirectShowCamera;

// Get a empty camera
UVCCamera camera = UVCCamera();

// Get available camera list. You can use getResolutions() to acquire the resolutions
std::vector<CameraDevice> cameraDeivceList = camera.getCameras();

// Open the first camera
camera.open(cameraDeivceList[0]);

// Start Capture
camera.startCapture();

// Wait 1 second
std::this_thread::sleep_for(std::chrono::seconds(1));

// Capture a frame
cv::Mat frame = camera.getMat();

// Close the camera
camera.close();

For more details, please see example code , Documentation and Software Diagrams

Looper

Looper is a loop thread worked in detach mode, making the camera life cycle easy to handle.

Usage

#include <uvc_camera_looper.h>
#include <iostream>

using namespace DirectShowCamera;

// Create a looper
UVCCameraLooper cameraLooper = UVCCameraLooper();

// Open the first camera
std::vector<CameraDevice> cameraDeivceList = cameraLooper.getCamera()->getCameras();
cameraLooper.getCamera()->open(cameraDeivceList[0]);

// Set a process to handle the captured image
cameraLooper.setCapturedProcess(
    [](cv::Mat image)
    {
        std::cout << "Get a new image" << std::endl;
    }
);

// Start looper to capture image continously
cameraLooper.start();

// 1 second
std::this_thread::sleep_for(std::chrono::seconds(1));

// Stop looper
cameraLooper.stop();

Exposure Fusion

Exposure fusion is a technique to capture a high quality image by fusing multi-exposure sequence.

cv::Mat frame = camera.exposureFusion();

Reference: Tom Mertens, Jan Kautz, and Frank Van Reeth. Exposure fusion. In Computer Graphics and Applications, 2007. PG'07. 15th Pacific Conference on, pages 382–390. IEEE, 2007.

Using camera without OpenCV

This library can be used without opencv.

1. Find ds_libs_setting.h in the include/directshow_camera folder.
2. Comment this line

   //#define HAS_OPENCV

Then you can use the following code to get frame bytes in BGR order.

int frameSize;
unsigned char* image = new unsigned char[1208 * 760 * 3];
camera.getFrame(&image, &size);

Requirements

Minimum C++ 17

vcpkg commit version: #14121

OpenCV version: 4.3.0 (Option)

Installation

1. Install Window SDK

1. Install: https://developer.microsoft.com/en-us/windows/downloads/windows-10-sdk/

2.1 Cmake

1. Create libs folder in you project

2. Clone this repository into the libs folder

   git clone https://github.com/kcwongjoe/directshow_camera.git
   

3. Add the code in your project CMakeLists.txt

   # Set Window SDK Path. Change it to your Window SDK Path
    set(CMAKE_WINDOWS_KITS_10_DIR "C:/Program Files (x86)/Windows Kits/10") 
   
   # Camera library
    add_subdirectory(./libs/directshow_camera)
    target_link_libraries(${PROJECT_NAME}
        PRIVATE
            directshow_camera
    )

4. Install OpenCV library in CMake from vcpgk (Option)

   1. Add the code in your project CMakeLists.txt

      # Set vcpkg path. Change here if the vcpkg folder is not located in your solution folder.
      set(VCPKG_PATH ${CMAKE_CURRENT_SOURCE_DIR}/vcpkg)
      
      set(CMAKE_TOOLCHAIN_FILE
          ${VCPKG_PATH}/scripts/buildsystems/vcpkg.cmake
          CACHE STRING "Vcpkg toolchain file")
      
      # OpenCV
      
      #   Install OpenCV from vcpkg. See cmake/InstallVcpkgOpenCV.cmake
      install_vcpkg_opencv()
      
      #   Link OpenCV
      include_directories( ${OpenCV_INCLUDE_DIRS} )
      link_directories( ${OpenCV_LIB_DIR} )
      
      target_link_libraries(${PROJECT_NAME}
          PRIVATE
              OpenCV_LIBS
      )

   2. Copy dll from vcpkg. You can use copy_dll.bat or build-in CMake module to copy necessary dll from vcpkg.

      • Type copy_dll.bat x86 or copy_dll.bat x64
      • Add the code in your project CMakeLists.txt

        # dll will be copied when the target built.
        prebuild_copy_vcpkg_opencv_dll("<YOUR_TARGET>" "<COPY TO, e.g. ${CMAKE_BINARY_DIR}/examples>")

2.2 For dummy

1. Copy all files inside src folder and include/directshow_camera folder to your project.

Deploy example codes

1. Install Window SDK

   1. Install: https://developer.microsoft.com/en-us/windows/downloads/windows-10-sdk/
   2. Go to config folder, Rename window_sdk_path_example.txt to window_sdk_path.txt and type your Window SDK path in the text file.

2. Clone this repository

   git clone --recurse-submodules https://github.com/kcwongjoe/directshow_camera.git
   

3. Install vcpkg in the project folder

   1. Clone the vcpkg repo from https://github.com/Microsoft/vcpkg
   2. Unzip and rename folder as vcpkg
   3. Place in the solution folder
   4. Run bootstrap-vcpkg.bat in the vcpkg folder
   5. If your vcpkg folder was not located in the solution folder, you have to specify the path in the config folder. Go to config folder, Rename vcpkg_path_example.txt to vcpkg_path.txt and type your vcpkg path in the text file.

4. Install OpenCV by vcpkg

   1. Go to vcpkg folder
   2. Type vcpkg install opencv:x86-windows
   3. Type vcpkg install opencv:x64-windows

5. Run build.bat in Solution folder

   Type build x86 or build x64

6. Go to build folder and open visual studio solution.

7. Build solution in visual studio. Executable file will be constructed but you may encounter a missing dll error. Then, go to Solution folder and run copy_dll.bat to copy dll from vcpkg folder.

   Type copy_dll x86 or copy_dll x64

License

This project is licensed under MIT license.