VisionMaster4.4 python脚本 图像处理 转换函数 爱之初体验

最近有接触过一丢丢VM4.3的模块开发. 一直有把python图像处理部分模块移植进来的打算 不过时间不够没来得及折腾.偶尔发现4.4支持py脚本 于是拿来折腾.一下午.

发现4.4支持python脚本,好开心.

首先安装VM4.4 注意一定要是4.4

打开后拖了一个模块.

但是发现import numpy import cv2 都报错.

python没有这些怎么玩? 咱好歹自学过几天python 不能放弃吧?

于是抱着试一试的态度?

开始以为路径是

C:\Program Files\VisionMaster4.4.0\Applications\Module(sp)\x64\Logic\PyShellModule

后来搜索感觉是这里:

C:\Program Files\VisionMaster4.4.0\Applications\ModuleProxy\x64python.exeC:\Program Files\VisionMaster4.4.0\Applications\ModuleProxy\x64\Scripts
pip.exe

补充一下:

C:\Program Files\VisionMaster4.4.0\Applications\ModuleProxy\x64\Scripts

这个目录是空的 需要双击执行get-pip.py等待文件下载完毕 里面才会出现 pip.exe等文件

好了,在这里按住shift右击 >>在此处打开powershell窗口 输入cmd

然后 尝试性的在此文件夹里 pip install numpy 当运行 import numpy as np 发现已经不报错了.

好了 现在直接在这个窗口:安装

pip install numpy
pip install opencv-python
pip install opencv-contrib-python
pip install pillow
pip install matplotlibpip install numpy pillow opencv-python opencv-contrib-python matplotlib
# 我暂时只需要这么多 够用 你也可以安装别的

当然halcon装不了:强制安装也会报错:如下所示:

C:\Program Files\VisionMaster4.4.0\Applications\ModuleProxy\x64\Scripts>pip install mvtec-halcon==23050
ERROR: Ignored the following versions that require a different python version: 20110.0.0 Requires-Python >=3.8; 20110.0.1 Requires-Python >=3.8; 20111.0.0 Requires-Python >=3.8; 20111.0.1 Requires-Python >=3.8; 20112.0.0 Requires-Python >=3.8; 20113.0.0 Requires-Python >=3.8; 20114.0.0 Requires-Python >=3.8; 21050.0.0 Requires-Python >=3.8; 21110.0.0 Requires-Python >=3.8; 22050.0.0 Requires-Python >=3.8; 22110.0.0 Requires-Python >=3.8; 22110.0.1 Requires-Python >=3.8; 22111.0.0 Requires-Python >=3.8; 22111.0.1 Requires-Python >=3.8; 22112.0.0 Requires-Python >=3.8; 22112.0.1 Requires-Python >=3.8; 22113.0.1 Requires-Python >=3.8; 23050.0.0 Requires-Python >=3.8; 23110.0.0 Requires-Python >=3.8; 23110.0.1 Requires-Python >=3.8; 24050.0.0 Requires-Python >=3.8; 24111.0.0 Requires-Python >=3.8
ERROR: Could not find a version that satisfies the requirement mvtec-halcon==23050 (from versions: none)
ERROR: No matching distribution found for mvtec-halcon==23050

意味着必须3.8才能安装python版的halcon实在是非常遗憾. 官方能不能把你python版本升级到3.8呢?????????? 卡一个版本 气死人.    这个后面再想想别的办法.试试看能不能补救. 实在不行只能pythonnet 用halcon的.net接口了. 或者加强反馈等官方更新.

哈哈:

接下来又遇到难题:

官方文档只介绍输入输出图像的结构体: 没有示例程序.难搞.

结构体定义如下:

# 图像数据
class ImageData:def __init__(self):self.width = Noneself.height = Noneself.pixel_format = Noneself.buffer = Noneself.dataLen = None

其他类型 数据结构比较简单 见名知意: 好吧还是放出来吧:

# 路径 文件名
#C:\Program Files\VisionMaster4.4.0\Applications\Module(sp)\x64\Logic\PyShellModule
#组合数据类型字段定义.py# ROI圆环
class RoiAnnulus:def __init__(self):self.center_x = Noneself.center_y = Noneself.inner_radius = Noneself.outer_radius = Noneself.start_angle = Noneself.angle_extend = None# 圆
class Circle:def __init__(self):self.radius = Noneself.center_x = Noneself.center_y = None# 椭圆
class ELLIPSE:def __init__(self):self.center_x = Noneself.center_y = Noneself.major_radius = Noneself.minor_radius = Noneself.angle = None# 位置修正信息
class Fixture:def __init__(self):self.init_point_x = Noneself.init_point_y = Noneself.init_angle = Noneself.init_scale_x = Noneself.init_scale_y = Noneself.run_point_x = Noneself.run_point_y = Noneself.run_angle = Noneself.run_scale_x = Noneself.run_scale_y = None# 图像数据
class ImageData:def __init__(self):self.width = Noneself.height = Noneself.pixel_format = Noneself.buffer = Noneself.dataLen = None# 直线
class Line:def __init__(self):self.start_point_x = Noneself.start_point_y = Noneself.end_point_x = Noneself.end_point_y = None# 点
class Point:def __init__(self):self.point_x = Noneself.point_y = None# 轮廓
class PointSet:def __init__(self):self.buffer = Noneself.dataLen = None# 矩形
class Rect:def __init__(self):self.rect_x = Noneself.rect_y = Noneself.width = Noneself.height = None# Box
class RoiBox:def __init__(self):self.center_x = Noneself.center_y = Noneself.width = Noneself.height = Noneself.angle = None# 多边形
class RoiPolygon:def __init__(self):self.point_num = Noneself.point_x = Noneself.point_y = None# 圆环
class Annulus:def __init__(self):self.center_x = Noneself.center_y = Noneself.inner_radius = Noneself.outer_radius = Noneself.start_angle = Noneself.angle_extend = None

直接上程序: 自己摸索的 输入输出 官方可没有示例 噢

# coding: utf-8
from ioHelper import *# pip install opencv-python
# pip install numpyimport os,sys,time
import cv2
import numpy as npdef Imagetonumpy(ImageData):# 将二进制数据转换为 NumPy 数组image_array = np.frombuffer(ImageData.buffer, dtype=np.uint8)# 调整数组形状为 (height, width, channels) # 图是3 通道 就选3 1通道就选1if (int(len(ImageData.buffer)) == int(ImageData.width*ImageData.height)):channels=1else:channels=3image_array = image_array.reshape((ImageData.height, ImageData.width,channels))if channels==3:image_array = cv2.cvtColor(image_array, cv2.COLOR_RGB2BGR)return image_array
#*******************************************************************************************
def numpytoImage(numpy_image):"""将 NumPy 数组转换为 ImageData 结构体。参数:numpy_image: NumPy 数组，形状为 (height, width, channels) 或 (height, width)。返回:ImageData: 包含图像数据的结构体。"""# 检查 NumPy 数组的形状if len(numpy_image.shape) == 1:  # 灰度图像height, width,channels = numpy_image.shapechannels = 1elif len(numpy_image.shape) == 3:  # 彩色图像height, width, channels = numpy_image.shapeelse:raise ValueError("NumPy 数组形状不正确！")if channels == 3:numpy_image = cv2.cvtColor(numpy_image, cv2.COLOR_RGB2BGR)buffer = numpy_image.tobytes()# 将 NumPy 数组转换为二进制数据# 创建 ImageData 结构体image_data = ImageData()image_data.width = widthimage_data.height = heightif channels == 3:image_data.pixel_format =  35127316 # 假设像素格式与原函数一致 35127316 if channels == 1:    image_data.pixel_format = 17301505  # 假设像素格式与原函数一致 35127316 image_data.buffer = bufferimage_data.dataLen = width*height*channels#image_data.pixel_format 必须正确 否则图像报错  我也不知带这个是啥格式啥结构 # 但是可以通过 输入图像 tmpimg.pixel_format 获取 这个数值#localVar = IoHelper(data, INIT_LOCAL_VAR)#tmpimg = moduleVar.img #读取图像#{tmpimg.width}// 宽度 759#{tmpimg.height}//高度 658#{tmpimg.pixel_format}return image_data   def Process(data) -> int:"""Write custom logic code inside the Process method. Return 0 for success or any other value for failure.""""""在Process方法内编写自定义逻辑代码, 成功返回0, 返回其他值表示失败""""""Do not delete this code block.""""""请勿删除此处代码"""moduleVar = IoHelper(data, INIT_MODULE_VAR)globalVar = IoHelper(data, INIT_GLOBAL_VAR)localVar = IoHelper(data, INIT_LOCAL_VAR)tmpimg = moduleVar.img #读取图像moduleVar.outstr = fr"""{tmpimg.width}// 宽度 759{tmpimg.height}//高度 658{tmpimg.pixel_format}//像素格式 //35127316[tmpimg.buffer]//图像数据  二进制{tmpimg.dataLen}//数据长度 //1498266{os.getcwd()}{int(len(tmpimg.buffer))}{int(tmpimg.width*tmpimg.height)}{int(len(tmpimg.buffer)) == int(tmpimg.width*tmpimg.height)}"""image_array=Imagetonumpy(tmpimg)#image = cv2.imread(tmpimg)# 使用 OpenCV 显示图像#blank_image = np.ones((200, 200, 1), dtype=np.uint8) * 255#numpy_image = np.zeros((100,100,3),dtype=np.uint8)#cv2.imshow("Image", image_array)#cv2.waitKey(2000)#cv2.destroyAllWindows()imgooop=numpytoImage(image_array)moduleVar.imgout=imgooop"""Variable initialization, to access input and output variables within the Process method.""""""变量初始化, 进入Process方法后首先调用, 否则无法访问输入、输出变量""""""* Input and output variables:* 输入、输出变量Input variables are read-only and cannot be modified; output variables are write-only and cannot be read.输入变量为只读, 不可以修改; 输出变量为只写, 不可以读取;Assume: the name of the input variable is 'in0', of type int. Its value can be obtained using tmp = moduleVar.in0假设: 输入变量名为in0, 类型为int, 可通过tmp = moduleVar.in0 读取该变量的值Assume: the name of the output variable is 'out0', of type int. Its value can be set using moduleVar.out = 9假设: 输出变量名为out0, 类型为int, moduleVar.out = 9 设置该变量的值Global variables:* 全局变量Assume: the name of the global variable is 'var0', of type int. Its value can be obtained using tmp = globalVar.GetValue('var0')假设: 输入变量名为var0, 类型为int, 可通过tmp = globalVar.GetValue('var0') 读取该变量的值, 通过 globalVar.SetValue('var0', 123) 设置该变量的值Local variables:* 局部变量Assume: the name of the local variable is 'var0', of type int. Its value can be obtained using tmp = localVar.GetValue('var0')假设: 输入变量名为var0, 类型为int, 可通过tmp = localVar.GetValue('var0') 读取该变量的值, 通过 localVar.SetValue('var0', 123) 设置该变量的值** All variables should be used strictly based on their types. If there is a need for type conversion, explicit conversion should be performed in the user code.** 所有变量都需要严格根据变量类型使用, 若有类型转换的需求, 应在用户代码内通过显示转换代码进行类型转换。Example code:示例代码如下:width = moduleVar.width + 100moduleVar.out0 = width / 2 + 20if moduleVar.in1 == 'OK':moduleVar.out1 = "OK"moduleVar.out0 = globalVar.GetValue('var0')moduleVar.out1 = localVar.GetValue('var0')globalVar.SetValue('var1', 123)localVar.SetValue('var1', moduleVar.out2)Note: Python indentation must be 4 spaces注意: Python的缩进必须为4个空格。编写自定义代码，替换下方的默认代码 passWrite custom code to replace the default code below.""""try:#PrintMsg("\nUser code start")pass#PrintMsg("User code end")except BaseException as e:PrintMsg(e)return 0

 仔细看 就是2个转换函数: process是被调用方

注意 tmpimg.pixel_format 我没搞懂啥格式.就是一串数字,但是 输入图像是多少,输出就写多少就没有问题.

偶然看文档发现:

file:///C:/Program%20Files/VisionMaster4.4.0/Applications/Help/CH/index.html
浏览 帮助文档 看到了介绍:
帮助文档里面>颜色处理>彩图生成>输出图像像素格式:
int型，代表输出图像的像素格式。其中17301505对应Mono8格式，35127316对应RGB24格式。
那么 意味着 程序 可以没有问题.

美中不足的是VM 用的 python3.7 64位 如果是python3.8 那么python 版本的halcon也可以支持