python PNP 根据点云三维坐标和图像二维坐标，计算雷达相机外参，并作重投影验证

最新推荐文章于 2024-04-06 09:48:14 发布

小白菜的点滴成长

最新推荐文章于 2024-04-06 09:48:14 发布

阅读量3.2k

点赞数 1

分类专栏：自动驾驶文章标签： opencv pnp 3d 2d

本文链接：https://blog.csdn.net/qq_18894441/article/details/102676480

版权

自动驾驶专栏收录该内容

24 篇文章 1 订阅

订阅专栏

#!/usr/bin/env python 
import cv2
import numpy as np 
import math
point3s=np.array(([-3.42315, -1.69773, -0.340899],[-3.76219, -1.10685, -0.341598],[-4.15474, -1.106, -1.40024],[-3.14925, -1.54094, -1.06652]),dtype=np.double)
point2s=np.array(([494, 89],[1028, 118],[1074, 754],[505, 764]),dtype=np.double)

camera=np.array(([ 1.8662051087073896e+03, 0, 9.6197416454354118e+02],[0,1.8907532330116028e+03, 6.8753441160022533e+02],[0,0,1]),dtype=np.double)
dist=np.array(([-5.7021027699312055e-01, 5.3788676876655150e-01,-1.3317400379136706e-02, 9.6858782674096261e-04,-5.2633103189649655e-01]),dtype=np.double)
#dist=dist.T
#dist=np.zeros((5,1))
found,r,t=cv2.solvePnP(point3s,point2s,camera,dist) #计算雷达相机外参,r-旋转向量，t-平移向量
#R=cv2.Rodrigues(r)[0] #旋转向量转旋转矩阵
#camera_position=-np.matrix(R).T*np.matrix(t) #相机位置

d3=np.array([[-3.14925, -1.54094, -1.06652]])
d2,_=cv2.projectPoints(d3,r,t,camera,dist)#重投影验证
print(r)
print(t)
print(d2)