c语言和python的数组-python与C语言之间的数组传递

参考文章：https://cloud.tencent.com/developer/news/333692

在Ubuntu系统下测试python调用C++程序，完成数组传入、传出、图片传输。其实图片就是特定数据类型的数组。

test.cpp:

#include "opencv2/opencv.hpp"

#define DLLEXPORT extern "C"

DLLEXPORT void sum_f(float* a,float* b,float* res,int n)

{

for(int i=0; i

动态库编译

g++ test.cpp -fpic -shared -o test.so `pkg-config --cflags --libs opencv`

单引号为调用默认安装的opencv

test.py

from ctypes import cdll

import numpy as np

import ctypes

#导入动态库

mydll = cdll.LoadLibrary('test.so')

# 函数声明

sum_f = mydll.sum_f

sum_f.argtypes = [np.ctypeslib.ndpointer(dtype=np.float32,ndim=1,flags="C_CONTIGUOUS"),

np.ctypeslib.ndpointer(dtype=np.float32,ndim=1,flags="C_CONTIGUOUS"),

np.ctypeslib.ndpointer(dtype=np.float32,ndim=1,flags="C_CONTIGUOUS"),

ctypes.c_int]

x = np.ones(12,dtype=np.float32)*2

y = np.ones(12,dtype=np.float32)*3

print('x=',x)

print('y=',y)

res = np.zeros(12,dtype=np.float32)

#函数调用

sum_f(x,y,res,12)

print('x+y=',res)

# test opencv ===================================

test_opencv = mydll.test_opencv

test_opencv.argtypes = [np.ctypeslib.ndpointer(dtype=np.uint8,ndim=1,flags="C_CONTIGUOUS"),

ctypes.c_int]

n = 255

img = np.zeros((n,n,3),np.uint8)

img[:,:,0] = 255

img = img.reshape(-1)

test_opencv(img,n)

输入为x、y，输出为res，为两个之和。