一、前言

最近来触碰一下halcon，一直以来作为ai算法工程师，虽然知道halcon，但是一直也没有用过

对于我们用户来说，halcon与opencv的差距主要在下面：

（1）halcon是闭源的，商业的软件，优点是算子丰富，应用成熟

（2）opencv是开源的，社区强大

我们只是使用者而已，很多公司也是有一个专门将halcon的功能替换为opencv的这么一个岗位，这样摆脱商业库的依赖。当然还有更牛逼的全部算子手写，不使用opencv。很明显我不是这种人！

二、Halcon例子实操

那好，我们先来看我们的第一个例子吧：

这个例子是来源于：

明显是个2d测量类的例子，主要是测量孔洞的平均半径和方差半径。

Halcon代码如下：

read_image (Image, 'progres')
get_image_size (Image, Width, Height)
dev_close_window ()
dev_open_window (0, 0, Width, Height, 'white', WindowID)
dev_set_color ('red')
gen_rectangle1 (Rectangle, 260, 90, 360, 350)
reduce_domain (Image, Rectangle, ImageReduced)
threshold (ImageReduced, Dark, 0, 150)
connection (Dark, ConnectedRegions)
select_shape (ConnectedRegions, SelectedRegions, 'area', 'and', 10, 100)
elliptic_axis (SelectedRegions, Ra, Rb, Phi)
area_center (SelectedRegions, Area, Row, Column)
dev_set_draw ('margin')
dev_display (Image)
* dev_set_colored (12)
dev_set_color ('green')
dev_display (SelectedRegions)
MeanRadius := sum(Ra) / |Ra|
VarianceRadius := sum(Ra * Ra - MeanRadius * MeanRadius) / (|Ra| - 1)

我们先来解读一下以上的代码：

1）read_image (Image, 'progres')：读取图像

2）get_image_size (Image, Width, Height)：获取读入图像的宽和高

3）dev_close_window ()：关闭窗口，放在这里也是为了关闭一些潜在的没有关闭的可视化窗口

4）dev_open_window (0, 0, Width, Height, 'white', WindowID)：打开一个窗口

5）dev_set_color ('red')：设置颜色

6）gen_rectangle1 (Rectangle, 260, 90, 360, 350)：生成一个矩形

7）reduce_domain(Image, Rectangle, ImageReduced)：　通过reduce_domain确实能获得特定区域Region位置的图像，但是，reduce_domain是缩小一个图像的定义域，并不缩小图像的实际尺寸，即新图像ImageReduced尺寸大小并未发生变化。如果使用get_image_size来计算ImageReduced图像的尺寸，其尺寸和原图Image一样。并且，包括一系列的domain算子，比如change_domain，均不改变图像矩阵的大小。

reduce_domain (Image, Rectangle, ImageReduced)：可以理解为，获取Image图上Rectangle区域大小的图，这个小图就是返回给ImageReduced。为了减小处理的区域

8）threshold (ImageReduced, Dark, 0, 150)：就是一个阈值分割，输入ImageReduced，输出

Dark。

9）connection (Dark, ConnectedRegions):连通域分析

10）select_shape (ConnectedRegions, SelectedRegions, 'area', 'and', 10, 100)：应该是根据面积和大小挑选。

11）elliptic_axis (SelectedRegions, Ra, Rb, Phi)：

12）area_center (SelectedRegions, Area, Row, Column)：

13）dev_set_draw ('margin')：

14）dev_display (Image)：显示图像Image，这个是原图

15）dev_set_color ('green')：设置后面待显示的对象的颜色

16）dev_display (SelectedRegions)：显示选择出来的区域，颜色就是绿色的

17）MeanRadius := sum(Ra) / |Ra|、VarianceRadius := sum(Ra * Ra - MeanRadius * MeanRadius) / (|Ra| - 1)：计算出所需要的测量值

可以看到其中的图像变量有多个图，但注意这些都是变量，直接halcon自己在变量窗口中自己显示出来的。

5张图的前三张都可以双击后，在画布上显示，分别按照顺序是：

获得的dark，可以看到没经过连通域分析，这时还有右下角一大块干扰。

剩下这两个图像变量就无法在画布上显示了（这个我也不晓得，刚接触halcon）

三、导出到C++

Halcon C++或者Halcon C#是很频繁的用法，想要将halcon落地到真正的应用上，这个是必须的。

那么如何导出到c++呢？请看下面的做法！

3.1、选择界面，文件-》导出语言...

导出代码如下：

///
// File generated by HDevelop for HALCON/C++ Version 21.11.0.0
// Non-ASCII strings in this file are encoded in local-8-bit encoding (cp936).
// Ensure that the interface encoding is set to locale encoding by calling
// SetHcppInterfaceStringEncodingIsUtf8(false) at the beginning of the program.
// 
// Please note that non-ASCII characters in string constants are exported
// as octal codes in order to guarantee that the strings are correctly
// created on all systems, independent on any compiler settings.
// 
// Source files with different encoding should not be mixed in one project.
///



#ifndef __APPLE__
#  include "HalconCpp.h"
#  include "HDevThread.h"
#else
#  ifndef HC_LARGE_IMAGES
#    include <HALCONCpp/HalconCpp.h>
#    include <HALCONCpp/HDevThread.h>
#    include <HALCON/HpThread.h>
#  else
#    include <HALCONCppxl/HalconCpp.h>
#    include <HALCONCppxl/HDevThread.h>
#    include <HALCONxl/HpThread.h>
#  endif
#  include <stdio.h>
#  include <CoreFoundation/CFRunLoop.h>
#endif



using namespace HalconCpp;


#ifndef NO_EXPORT_MAIN
// Main procedure 
void action()
{

  // Local iconic variables
  HObject  ho_Image, ho_Rectangle, ho_ImageReduced;
  HObject  ho_Dark, ho_ConnectedRegions, ho_SelectedRegions;

  // Local control variables
  HTuple  hv_Width, hv_Height, hv_WindowID, hv_Ra;
  HTuple  hv_Rb, hv_Phi, hv_Area, hv_Row, hv_Column, hv_MeanRadius;
  HTuple  hv_VarianceRadius;

  ReadImage(&ho_Image, "progres");
  GetImageSize(ho_Image, &hv_Width, &hv_Height);
  if (HDevWindowStack::IsOpen())
    CloseWindow(HDevWindowStack::Pop());
  SetWindowAttr("background_color","white");
  OpenWindow(0,0,hv_Width,hv_Height,0,"visible","",&hv_WindowID);
  HDevWindowStack::Push(hv_WindowID);
  if (HDevWindowStack::IsOpen())
    SetColor(HDevWindowStack::GetActive(),"red");
  GenRectangle1(&ho_Rectangle, 260, 90, 360, 350);
  ReduceDomain(ho_Image, ho_Rectangle, &ho_ImageReduced);
  Threshold(ho_ImageReduced, &ho_Dark, 0, 150);
  Connection(ho_Dark, &ho_ConnectedRegions);
  SelectShape(ho_ConnectedRegions, &ho_SelectedRegions, "area", "and", 10, 100);
  EllipticAxis(ho_SelectedRegions, &hv_Ra, &hv_Rb, &hv_Phi);
  AreaCenter(ho_SelectedRegions, &hv_Area, &hv_Row, &hv_Column);
  if (HDevWindowStack::IsOpen())
    SetDraw(HDevWindowStack::GetActive(),"margin");
  if (HDevWindowStack::IsOpen())
    DispObj(ho_Image, HDevWindowStack::GetActive());
  //dev_set_colored (12)
  if (HDevWindowStack::IsOpen())
    SetColor(HDevWindowStack::GetActive(),"green");
  if (HDevWindowStack::IsOpen())
    DispObj(ho_SelectedRegions, HDevWindowStack::GetActive());
  hv_MeanRadius = (hv_Ra.TupleSum())/(hv_Ra.TupleLength());
  hv_VarianceRadius = (((hv_Ra*hv_Ra)-(hv_MeanRadius*hv_MeanRadius)).TupleSum())/((hv_Ra.TupleLength())-1);
}


#ifndef NO_EXPORT_APP_MAIN

#ifdef __APPLE__
// On OS X systems, we must have a CFRunLoop running on the main thread in
// order for the HALCON graphics operators to work correctly, and run the
// action function in a separate thread. A CFRunLoopTimer is used to make sure
// the action function is not called before the CFRunLoop is running.
// Note that starting with macOS 10.12, the run loop may be stopped when a
// window is closed, so we need to put the call to CFRunLoopRun() into a loop
// of its own.
static HMutex*     sStartMutex;
static H_pthread_t sActionThread;
static bool        sTerminate = false;

static void timer_callback(CFRunLoopTimerRef timer, void *info)
{
  sStartMutex->UnlockMutex();
}

static Herror apple_action(void **parameters)
{
  // Wait until the timer has fired to start processing.
  sStartMutex->LockMutex();
  sStartMutex->UnlockMutex();

  try
  {
    action();
  }
  catch (HException &exception)
  {
    fprintf(stderr,"  Error #%u in %s: %s\n", exception.ErrorCode(),
            exception.ProcName().TextA(),
            exception.ErrorMessage().TextA());
  }

  // Tell the main thread to terminate itself.
  sStartMutex->LockMutex();
  sTerminate = true;
  sStartMutex->UnlockMutex();
  CFRunLoopStop(CFRunLoopGetMain());
  return H_MSG_OK;
}

static int apple_main(int argc, char *argv[])
{
  Herror                error;
  CFRunLoopTimerRef     Timer;
  CFRunLoopTimerContext TimerContext = { 0, 0, 0, 0, 0 };

  sStartMutex = new HMutex("type","sleep");
  sStartMutex->LockMutex();

  error = HpThreadHandleAlloc(&sActionThread);
  if (H_MSG_OK != error)
  {
    fprintf(stderr,"HpThreadHandleAlloc failed: %d\n", error);
    exit(1);
  }

  error = HpThreadCreate(sActionThread,0,apple_action);
  if (H_MSG_OK != error)
  {
    fprintf(stderr,"HpThreadCreate failed: %d\n", error);
    exit(1);
  }

  Timer = CFRunLoopTimerCreate(kCFAllocatorDefault,
                               CFAbsoluteTimeGetCurrent(),0,0,0,
                               timer_callback,&TimerContext);
  if (!Timer)
  {
    fprintf(stderr,"CFRunLoopTimerCreate failed\n");
    exit(1);
  }
  CFRunLoopAddTimer(CFRunLoopGetCurrent(),Timer,kCFRunLoopCommonModes);

  for (;;)
  {
    bool terminate;

    CFRunLoopRun();

    sStartMutex->LockMutex();
    terminate = sTerminate;
    sStartMutex->UnlockMutex();

    if (terminate)
      break;
  }

  CFRunLoopRemoveTimer(CFRunLoopGetCurrent(),Timer,kCFRunLoopCommonModes);
  CFRelease(Timer);

  error = HpThreadHandleFree(sActionThread);
  if (H_MSG_OK != error)
  {
    fprintf(stderr,"HpThreadHandleFree failed: %d\n", error);
    exit(1);
  }

  delete sStartMutex;
  return 0;
}
#endif

int main(int argc, char *argv[])
{
  int ret = 0;

  try
  {
#if defined(_WIN32)
    SetSystem("use_window_thread", "true");
#endif

    // file was stored with local-8-bit encoding
    //   -> set the interface encoding accordingly
    SetHcppInterfaceStringEncodingIsUtf8(false);

    // Default settings used in HDevelop (can be omitted)
    SetSystem("width", 512);
    SetSystem("height", 512);

#ifndef __APPLE__
    action();
#else
    ret = apple_main(argc,argv);
#endif
  }
  catch (HException &exception)
  {
    fprintf(stderr,"  Error #%u in %s: %s\n", exception.ErrorCode(),
            exception.ProcName().TextA(),
            exception.ErrorMessage().TextA());
    ret = 1;
  }
  return ret;
}

#endif


#endif

3.2、vs中运行

我们在vs2019中配置运行一下：

halcon.props：

<?xml version="1.0" encoding="utf-8"?>
<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
  <ImportGroup Label="PropertySheets" />
  <PropertyGroup Label="UserMacros" />
  <PropertyGroup>
        <IncludePath>E:\setup_app\opencv453\build\include;../../3dparty/halcon_2111/include;../../3dparty/halcon_2111/include/halconcpp;../../3dparty/halcon_2111/include/hlib;../../3dparty/halcon_2111/include/hdevengine;../../3dparty/halcon_2111/include/hclib;../../3dparty/halcon_2111/include/halconc;../../3dparty/glog/include;$(IncludePath)</IncludePath>

        <LibraryPath Condition="'$(Platform)'=='X64'">E:\setup_app\opencv453\build\x64\vc14\lib;../../3dparty/halcon_2111/x64-win64/lib;../../3dparty/glog/lib;$(LibraryPath)</LibraryPath>
        <ExecutablePath Condition="'$(Platform)'=='X64'">E:\setup_app\opencv453\build\x64\vc14\bin;$(ExecutablePath)</ExecutablePath>


  </PropertyGroup>
  <ItemDefinitionGroup>
        <Link Condition="'$(Configuration)'=='Debug'">
          <AdditionalDependencies>
			  opencv_world453d.lib;
			%(AdditionalDependencies)</AdditionalDependencies>
        </Link>
        <Link Condition="'$(Configuration)'=='Release'">
          <AdditionalDependencies>
			  opencv_world453.lib;
			  glog.lib;
			  halcon.lib;
			  halconc.lib;
			  halconcpp.lib;
			  halconcppxl.lib;
			  halconcxl.lib;
			  halconxl.lib;
			  hdevenginecpp.lib;
			  hdevenginecppxl.lib;
			  libiomp5md.lib;
			 %(AdditionalDependencies)</AdditionalDependencies>
        </Link>
  </ItemDefinitionGroup>
  <ItemGroup />
</Project>

为了让我们的显示窗口不要一闪而过，因为我们需要添加"等待函数"：

注意：这个还需要一个附加库目录，否则会报错！

可以看到，要加等待函数，其次要想获得halcon的变量的值，必须前面使用double等数值类型，才可以看到，否则是无法看到具体数值的，最后我这里先给出疑问，如何像Halcon中那样可以看到图像变量，并进行可视化呢？这个后面再回过头来解决！

3.3、精简原始导出的代码，并做代码解析

我一开始也是不明白导出的那么多东西，是如何精简。后面根据使用过的人说一般是用不到的直接删掉。

因此我小心删除，精简后的代码如下：


#include "HalconCpp.h"
#include <stdio.h>



using namespace HalconCpp;



// Main procedure 
void action()
{

    // Local iconic variables
    HObject  ho_Image, ho_Rectangle, ho_ImageReduced;
    HObject  ho_Dark, ho_ConnectedRegions, ho_SelectedRegions;

    // Local control variables
    HTuple  hv_Width, hv_Height, hv_WindowID, hv_Ra;
    HTuple  hv_Rb, hv_Phi, hv_Area, hv_Row, hv_Column, hv_MeanRadius;
    HTuple  hv_VarianceRadius;

    ReadImage(&ho_Image, "progres");
    GetImageSize(ho_Image, &hv_Width, &hv_Height);
    if (HDevWindowStack::IsOpen())
        CloseWindow(HDevWindowStack::Pop());
    SetWindowAttr("background_color", "white");
    OpenWindow(0, 0, hv_Width, hv_Height, 0, "visible", "", &hv_WindowID);
    HDevWindowStack::Push(hv_WindowID);
    if (HDevWindowStack::IsOpen())
        SetColor(HDevWindowStack::GetActive(), "red");
    GenRectangle1(&ho_Rectangle, 260, 90, 360, 350);
    ReduceDomain(ho_Image, ho_Rectangle, &ho_ImageReduced);
    Threshold(ho_ImageReduced, &ho_Dark, 0, 150);
    Connection(ho_Dark, &ho_ConnectedRegions);
    SelectShape(ho_ConnectedRegions, &ho_SelectedRegions, "area", "and", 10, 100);
    EllipticAxis(ho_SelectedRegions, &hv_Ra, &hv_Rb, &hv_Phi);
    AreaCenter(ho_SelectedRegions, &hv_Area, &hv_Row, &hv_Column);
    if (HDevWindowStack::IsOpen())
        SetDraw(HDevWindowStack::GetActive(), "margin");
    if (HDevWindowStack::IsOpen())
        DispObj(ho_Image, HDevWindowStack::GetActive());
        WaitSeconds(5.0);
    //dev_set_colored (12)
    if (HDevWindowStack::IsOpen())
        SetColor(HDevWindowStack::GetActive(), "green");
    if (HDevWindowStack::IsOpen())
        DispObj(ho_SelectedRegions, HDevWindowStack::GetActive());
        WaitSeconds(5.0);
        
    hv_MeanRadius = (hv_Ra.TupleSum()) / (hv_Ra.TupleLength());
    hv_VarianceRadius = (((hv_Ra * hv_Ra) - (hv_MeanRadius * hv_MeanRadius)).TupleSum()) / ((hv_Ra.TupleLength()) - 1);

    std::cout << "hv_MeanRadius平均半径:" << double(hv_MeanRadius) << std::endl;
    std::cout << "hv_VarianceRadius方差半径:" << double(hv_VarianceRadius) << std::endl;
}




int main(int argc, char* argv[])
{
    int ret = 0;

    try
    {
        // file was stored with local-8-bit encoding
        //   -> set the interface encoding accordingly
        SetHcppInterfaceStringEncodingIsUtf8(false);

        // Default settings used in HDevelop (can be omitted)
        SetSystem("width", 512);
        SetSystem("height", 512);

        action();
    }
    catch (HException& exception)
    {
        fprintf(stderr, "  Error #%u in %s: %s\n", exception.ErrorCode(),
            exception.ProcName().TextA(),
            exception.ErrorMessage().TextA());
        ret = 1;
    }
    return ret;
}

配置运行还是跟前面的是一样的。

那么我们做下代码解析吧

（1）先看主函数

整体上是一个异常捕捉try-catch结构，其中 SetHcppInterfaceStringEncodingIsUtf8(false);这个就是从Halcon v18.x开始支持中文路径，但是要事先调用 SetHcppInterfaceStringEncodingIsUtf8(false);算子来进行设置支持中文解析。

halcon 默认最大显示为512*512

（2）进入action函数

这个action函数实际才是我们的最重要的算法实现部分！

// Main procedure 
void action()
{

    // Local iconic variables
    HObject  ho_Image, ho_Rectangle, ho_ImageReduced;
    HObject  ho_Dark, ho_ConnectedRegions, ho_SelectedRegions;

    // Local control variables
    HTuple  hv_Width, hv_Height, hv_WindowID, hv_Ra;
    HTuple  hv_Rb, hv_Phi, hv_Area, hv_Row, hv_Column, hv_MeanRadius;
    HTuple  hv_VarianceRadius;

    ReadImage(&ho_Image, "progres");
    GetImageSize(ho_Image, &hv_Width, &hv_Height);
    if (HDevWindowStack::IsOpen())
        CloseWindow(HDevWindowStack::Pop());
    SetWindowAttr("background_color", "white");
    OpenWindow(0, 0, hv_Width, hv_Height, 0, "visible", "", &hv_WindowID);
    HDevWindowStack::Push(hv_WindowID);
    if (HDevWindowStack::IsOpen())
        SetColor(HDevWindowStack::GetActive(), "red");
    GenRectangle1(&ho_Rectangle, 260, 90, 360, 350);
    ReduceDomain(ho_Image, ho_Rectangle, &ho_ImageReduced);
    Threshold(ho_ImageReduced, &ho_Dark, 0, 150);
    Connection(ho_Dark, &ho_ConnectedRegions);
    SelectShape(ho_ConnectedRegions, &ho_SelectedRegions, "area", "and", 10, 100);
    EllipticAxis(ho_SelectedRegions, &hv_Ra, &hv_Rb, &hv_Phi);
    AreaCenter(ho_SelectedRegions, &hv_Area, &hv_Row, &hv_Column);
    if (HDevWindowStack::IsOpen())
        SetDraw(HDevWindowStack::GetActive(), "margin");
    if (HDevWindowStack::IsOpen())
        DispObj(ho_Image, HDevWindowStack::GetActive());
        WaitSeconds(5.0);
    //dev_set_colored (12)
    if (HDevWindowStack::IsOpen())
        SetColor(HDevWindowStack::GetActive(), "green");
    if (HDevWindowStack::IsOpen())
        DispObj(ho_SelectedRegions, HDevWindowStack::GetActive());
        WaitSeconds(5.0);
        
    hv_MeanRadius = (hv_Ra.TupleSum()) / (hv_Ra.TupleLength());
    hv_VarianceRadius = (((hv_Ra * hv_Ra) - (hv_MeanRadius * hv_MeanRadius)).TupleSum()) / ((hv_Ra.TupleLength()) - 1);

    std::cout << "hv_MeanRadius平均半径:" << double(hv_MeanRadius) << std::endl;
    std::cout << "hv_VarianceRadius方差半径:" << double(hv_VarianceRadius) << std::endl;
}

局部符号变量。

局部控制变量。主要是一些属性的表达，如图的宽高、半径、弧长等。

那么我们这里可以看到HObject和HTuple这两个类，那为什么分别用这两个类来声明变量呢？

答：HALCON里面有很多数据类型，但是通过的导出向导工具导出后，就剩下两种数据类型：Hobject HTuple。一个是图像一个是数据。

头文件有定义这两个C++类：

C:\Program Files\MVTec\HALCON-20.11-Progress\include\halconcpp\HObject.h

C:\Program Files\MVTec\HALCON-20.11-Progress\include\halconcpp\HTuple.h

HObject:

// Represents an instance of an iconic object(-array). Base class for images, regions and XLDs
class LIntExport HObject : public HObjectBase
{
 
public:
 
  // Create an uninitialized instance
  HObject():HObjectBase() {}
 
  // Copy constructor
  HObject(const HObject& source) : HObjectBase(source) {}
 
  // Create HObject from object id. For copy=false takes
  // over management of input key. Type of key must match!
  explicit HObject(Hkey key, bool copy=true);
 
  // Access of object tuple element
  const HObject operator [] (Hlong index) const;
 
  // Deep copy of all data represented by this object instance
  HObject Clone() const;
 
 
 
 
  /***************************************************************************
   * Operators                                                               *
   ***************************************************************************/
 
  // Calculate the difference of two object tuples.
  HObject ObjDiff(const HObject& ObjectsSub) const;
 
  // Convert an "integer number" into an iconic object.
  void IntegerToObj(const HTuple& SurrogateTuple);
 
  // Convert an "integer number" into an iconic object.
  void IntegerToObj(void* SurrogateTuple);
 
  // Convert an iconic object into an "integer number."
  HTuple ObjToInteger(Hlong Index, Hlong Number) const;
 
  // Copy an iconic object in the HALCON database.
  HObject CopyObj(Hlong Index, Hlong NumObj) const;
 
  // Concatenate two iconic object tuples.
  HObject ConcatObj(const HObject& Objects2) const;
 
  // Select objects from an object tuple.
  HObject SelectObj(const HTuple& Index) const;
 
  // Select objects from an object tuple.
  HObject SelectObj(Hlong Index) const;
 
  // Compare iconic objects regarding equality.
  Hlong CompareObj(const HObject& Objects2, const HTuple& Epsilon) const;
 
  // Compare iconic objects regarding equality.
  Hlong CompareObj(const HObject& Objects2, double Epsilon) const;
 
  // Compare image objects regarding equality.
  Hlong TestEqualObj(const HObject& Objects2) const;
 
  // Number of objects in a tuple.
  Hlong CountObj() const;
 
  // Informations about the components of an image object.
  HTuple GetChannelInfo(const HString& Request, const HTuple& Channel) const;
 
  // Informations about the components of an image object.
  HString GetChannelInfo(const HString& Request, Hlong Channel) const;
 
  // Informations about the components of an image object.
  HString GetChannelInfo(const char* Request, Hlong Channel) const;
 
#ifdef _WIN32
  // Informations about the components of an image object.
  HString GetChannelInfo(const wchar_t* Request, Hlong Channel) const;
#endif
 
  // Name of the class of an image object.
  HTuple GetObjClass() const;
 
  // Create an empty object tuple.
  void GenEmptyObj();
 
  // Displays image objects (image, region, XLD).
  void DispObj(const HWindow& WindowHandle) const;
 
  // Read an iconic object.
  void ReadObject(const HString& FileName);
 
  // Read an iconic object.
  void ReadObject(const char* FileName);
 
#ifdef _WIN32
  // Read an iconic object.
  void ReadObject(const wchar_t* FileName);
#endif
 
  // Write an iconic object.
  void WriteObject(const HString& FileName) const;
 
  // Write an iconic object.
  void WriteObject(const char* FileName) const;
 
#ifdef _WIN32
  // Write an iconic object.
  void WriteObject(const wchar_t* FileName) const;
#endif
 
  // Deserialize a serialized iconic object.
  void DeserializeObject(const HSerializedItem& SerializedItemHandle);
 
  // Serialize an iconic object.
  HSerializedItem SerializeObject() const;
 
  // Insert objects into an iconic object tuple.
  HObject InsertObj(const HObject& ObjectsInsert, Hlong Index) const;
 
  // Remove objects from an iconic object tuple.
  HObject RemoveObj(const HTuple& Index) const;
 
  // Remove objects from an iconic object tuple.
  HObject RemoveObj(Hlong Index) const;
 
  // Replaces one or more elements of an iconic object tuple.
  HObject ReplaceObj(const HObject& ObjectsReplace, const HTuple& Index) const;
 
  // Replaces one or more elements of an iconic object tuple.
  HObject ReplaceObj(const HObject& ObjectsReplace, Hlong Index) const;
 
private:
 
  // Verify matching semantic type ('object')!
  void AssertObjectClass();
 
};

红色框中的api：

1）ReadImage：读取图片，同Hdev

2）GetImageSize：获取图片的宽高，同Hdev

其余的api也是同Hdev中的一样功能。

这里要说明一下SelectShape、EllipticAxis和AreaCenter：

3）select_shape：

参考：Halcon 算子 select_shape_郑建广视觉的博客-CSDN博客_selectshape

select_shape(select_shape(Regions : SelectedRegions : Features, Operation, Min, Max : ))：算子select_shape根据形状特征选择区域，对于输入的每个区域，将计算指定的特征根据参数（features）。如果每个(Operation = 'and')或至少一个(Operation = 'or')的计算特性在默认限制(Min,Max)内，该区域将被输出。输入参数关系符（'and', 'or'）如果features只是用一种特性那么此参数没有意义。

Features的值：'area':区域的面积、'row':中心的行坐标、'column':中心的列坐标、'width':区域宽度（平行于坐标轴）、'height':区域的高度（平行于坐标轴）、'row1':区域左上角的行坐标、'circularity':和圆的相似度、'compactness':密度、'contlength':轮廓的总长度、'convexity':凸度、'rectangularity':类矩形因子、ra':主半径的等效椭圆、'rb':次半径的等效椭圆、'phi':和X轴的夹角、'anisometry:'等距、'bulkiness:'蓬松性、'dist_mean':从区域边界到区域中心的平均距离。海鸥很多。。。。

4）EllipticAxis：

参考:halcon-elliptic_axis计算区域的等效椭圆数据 - 天子骄龙 - 博客园

elliptic_axis (Region, Ra, Rb, Phi)
*作用：计算区域的等效椭圆数据
*参数1：输入区域
*参数2：输出主半径
*参数3：输出次半径
*参数4：输出主半径和X轴的夹角（以弧度为单位  - pi / 2 < Phi && Phi <= pi / 2）
*计算与输入区域具有相同方向和相同宽高比的椭圆的半径Ra和Rb以及方向Phi，多个输入区域可以作为数组传递

5）AreaCenter：

参考：https://www.cnblogs.com/liming19680104/p/15938000.html

area_center (Region, Area, Row, Column)
*返回区域的面积和中心位置
*参数1：输入对象，即被测区域或图像
*参数2：输出变量，即被测区域的面积
*参数3：输出变量，即被测区域中心的行索引
*参数4：输出变量，即被测区域中心的列索引

参考资料：

1、halcon readimage不支持中文路径？_利白的博客-CSDN博客_halcon write_image 中文路径

2、halcon C++编程第17讲 C++窗口函数（tcy）_tcy23456的博客-CSDN博客_dev_open_window

3、Halcon之可视化_yangyang_z的博客-CSDN博客_halcon中dump-window

4、HALCON 21.11：深度学习笔记---模型(8)_机器视觉001的博客-CSDN博客

5、b站上的

6、 Halcon学习--鱼棒长度测量--被测物体显示不全--去除方法_枫呱呱的博客-CSDN博客_halcon图像显示不全

7、Halcon之HALCON/C++ 接口的基础知识_yangyang_z的博客-CSDN博客_halcon c++

8、Halcon导出的C#代码的调用及讨论_HDevelop的c#dll-C#文档类资源-CSDN下载

9、Halcon的C++教程_爱上解放晚晚的博客-CSDN博客

10、halcon c++版本窗口不显示图_Angaoun的博客-CSDN博客_halcon c++代码不显示窗口

11、C++中显示Halcon窗口（2D&3D）_爱吃橙子的哈士奇的博客-CSDN博客_visualize_object_model_3d

12、halcon多目标模板匹配示例_halcon多模板匹配,halcon多模板匹配-C++文档类资源-CSDN下载

13、https://wang520.blog.csdn.net/article/details/108875243?spm=1001.2101.3001.6650.3&utm_medium=distribute.pc_relevant.none-task-blog-2%7Edefault%7ECTRLIST%7ERate-3-108875243-blog-122517477.pc_relevant_aa2&depth_1-utm_source=distribute.pc_relevant.none-task-blog-2%7Edefault%7ECTRLIST%7ERate-3-108875243-blog-122517477.pc_relevant_aa2&utm_relevant_index=6

14、VS2019配置Halcon的c++环境傻瓜式教程_asd_sz的博客-CSDN博客_vs2019配置halcon

15、Halcon缺陷检测实例转OpenCV实现(二) PCB印刷缺陷检测_51CTO博客_Halcon缺陷检测

16、2.2超人视觉初级班模板匹配（基于灰度）实战第二讲_哔哩哔哩_bilibili

17、VS2019配置Halcon的c++环境傻瓜式教程_asd_sz的博客-CSDN博客_vs2019配置halcon

18、Halcon学习（四）导出C++代码，在VS2010下编译_zxucver的博客-CSDN博客_halcon20.11导出c++

19、HDevelop图形窗口操作（下）-工业视觉/halcon-少有人走的路