OpenCV工作流SDK之模板匹配

OpenCV学堂

发布于 2026-04-02 21:39:53

810

工作流介绍

OpenCV工作流SDK介绍与下载地址

https://gitee.com/opencv_ai/opencv_workflow

C++SDK配置与演示

下载C++ SDK开发包，下载地址

https://gitee.com/opencv_ai/opencv_workflow

解压缩开发包

打开VS2022，配置包含目录

D:\vcworkspaces\oemts_workflow_engine_lib\include\oemts

配置库目录

D:\vcworkspaces\oemts_workflow_engine_lib\lib

直接编译提供的示例代码，然后把oemts_workflow_engine.dll 文件放到项目生成目录下即可运行exe，运行结果如下：

完整的示例代码如下：

特别提醒：

请先配置好OpenCV4.10 C++依赖。

边缘模板匹配演示

首先配置好开发环境，OpenCV工作流最新版本依赖于OpenCV4.10，所以先配置好OpenCV4.10 C++ 开发环境，然后配置好OpenCV工作流。首先加载工作流配置文件

// load opencv workflow sdk
std::shared_ptr<QTongCoreCVWorkFlow> engine(new QTongCoreCVWorkFlow());
bool succ = engine->initWorkFlow("D:/python/opencv_workflow/template_demo.vm", 
"ba8e9ea82fb37f1722ad2d9f4962e7e4");
if (!succ) {
  std::cout << "Could not load workflow file here..." << std::endl;
  return -1;
}

然后加载测试图像并显示

//  image view
cv::Mat frame = cv::imread("D:/images/vm_test/target.bmp");
cv::imshow("OpenCV学堂-模板匹配", frame);

调用查询模板匹配结果

// query matched boxes information
auto boxes = engine->getMatchedBoxInfos();

完整的源代码如下：

#include "main_workflow.h"
#include <iostream>
#include <fstream>"
int main(int argc, char** argv) {
    // load opencv workflow sdk
    std::shared_ptr<QTongCoreCVWorkFlow> engine(new QTongCoreCVWorkFlow());
    bool succ = engine->initWorkFlow("D:/python/opencv_workflow/my_test.vm", 
                                                "ba8e9ea82fb37f1722ad2d9f4962e7e4");
    if (!succ) {
        std::cout << "Could not load workflow file here..." << std::endl;
        return -1;
    }
    //  image view
    cv::Mat frame = cv::imread("D:/python/opencv_workflow/target.bmp");
    cv::namedWindow("OpenCV学堂-边缘匹配", cv::WINDOW_NORMAL);
    cv::imshow("OpenCV学堂-边缘匹配", frame);
    // measure with caliper sdk
    cv::Mat result;
    std::vector<std::string> logs;
    engine->run_workflow(frame, result, logs);
    cv::namedWindow("OpenCV学堂-边缘匹配结果", cv::WINDOW_NORMAL);
    cv::imshow("OpenCV学堂-边缘匹配结果", result);
    cv::waitKey(0);
    cv::destroyAllWindows();
    return 0;
}

运行结果如下：