- 计算需要用到的库
from PIL import Image
import matplotlib.pyplot as plt
import scipy.stats as sts
import numpy as np
import cv2 as cv
from numpy import cov,corrcoef
from scipy.special import comb, perm
from itertools import combinations, permutations
- 一维统计量
- mean 均值
- ptp 极差
- var 方差
- std 标准差
- min max 极值
代码如下:
img1 = np.array(cv.imread("X:/WDR/ETM/LE71220352017080EDC00_B1.TIF",1))
img2 = ....
img7 = ....
IMGN2P=[img1,img2,img3,img4,img5,img7,img8] // 二维矩阵数组
img1np=np.array(img1).flatten()
img2np=.....
img7np=......
IMGNP=[img1np,img2np,img3np,img4np,img5np,img7np,img8np] // 一维矩阵数组
// 创建 数组
mean=[0.00,1.00,2.00,3.00,4.00,5.00,6.00] # 均值数组
ptp=range(7) # 极差
var=range(7) # 方差
std=range(7)# 标准差
正文代码:
mean=np.array(mean) ### 均值矩阵
ptp=np.array(ptp,dtype=float) ### 极差矩阵
var=np.array(var,dtype=float) ### 方差矩阵
std=np.array(std,dtype=float) ### 标准差
for x in range(7):
mean[x]=IMGN2P[x].mean()
var[x]=IMGN2P[x].var()
std[x]=IMGN2P[x].std()
ptp[x]=IMGN2P[x].ptp()
for x in range(7): ### 最大值,最小值5
print("各波段 max: ",IMGNP[x].max())
print("各波段 min: ",IMGNP[x].min())
- 多维统计量
- cov 协方差
- corrcoef 相关系数
- OIF 指数因子
先写单一协方差和相关系数 还有OIF :
#cov corrcof ###### 协方差矩阵 矩阵
data1 = np.array([img1np,img2np]) ##波段1.2
cove = cov(data1,bias=1)
print("波段1 2 协方差:",cove)
print("波段1 2 相关系数: ",corrcoef(data1))
###1 2 3 波段OIF
Sk=std[0]+std[1]+std[2]
data2= np.array([img1np,img3np]) ## 波段 1 3
data3=np.array([img2np,img3np]) ##波段 2 3
R1=abs(corrcoef(data1).min())
R2=abs(corrcoef(data2).min())
R3=abs(corrcoef(data3).min())
R=R1+R2+R3
OIF=Sk/R
print("OIF(1 2 3 ) : \n",OIF)
7波段协方差,相关系数和OIF:
# ncb73=comb(7,3)
nstd=list(range(100))
# ncb72=comb(7,2)
ncorrcoef=range(21)
k=0
npcrroef=np.zeros((7,7),dtype=float)
for x in range(6): ###相关系数
mxk=x
for y in range(1,6-x):
mxk=mxk+1
ndata=np.array([IMGNP[x],IMGNP[mxk]]) ### ju zhen qiu xiang guan xi shu 相关系数矩阵 12 13 ....
npcrroef[x][mxk]=abs(corrcoef(ndata).min())
k=k+1
m2=0
mk=0
for x in range(5):
m2=m2+1
men=m2
for y in range(5-m2):
men=men+1
nstd[mk]=std[x]+std[m2]+std[men]
for x in range(35):
if x < 5:
nco = npcrroef[0][x+2]+npcrroef[0][1]+npcrroef[1][x+2]
OIF = nstd[x]/nco
print("OIF( 1 2 ",x+3," )",OIF)
elif x < 9:
nco = npcrroef[0][2]+npcrroef[0][x-2]+npcrroef[2][x-2]
OIF = nstd[x]/nco
print("OIF(1 3 ",x-1," )",OIF)
elif x < 12:
nco = npcrroef[0][3]+npcrroef[0][x-5]+npcrroef[3][x-5]
OIF = nstd[x]/nco
print("OIF ( 1 4 ",x-4," )",OIF)
elif x < 14:
nco = npcrroef[0][4]+npcrroef[0][x-7]+npcrroef[4][x-7]
OIF= nstd[x]/nco
print("OIF ( 1 5 ",x-7," )",OIF)
elif x <15:
nco = npcrroef[0][5]+npcrroef[0][6]+npcrroef[5][6]
OIF= nstd[x]/nco
print("OIF ( 1 6 7)",OIF)
elif x < 19:
nco = npcrroef[1][2]+npcrroef[1][x-12]+npcrroef[2][x-12]
OIF=nstd[x]/nco
print("OIF ( 2 3 ",x-11," )",OIF)
elif x < 22:
nco=npcrroef[1][3]+npcrroef[1][x-15]+npcrroef[3][x-15]
OIF=nstd[x]/nco
print("OIF ( 2 4 ",x-14," )",OIF)
elif x< 24:
nco=npcrroef[1][4]+npcrroef[1][x-17]+npcrroef[4][x-17]
OIF=nstd[x]/nco
print("OIF (2 5 ",x-16," )",OIF)
elif x <25:
nco=npcrroef[1][5]+npcrroef[1][x-18]+npcrroef[5][x-18]
OIF=nstd[x]/nco
print("OIF (2 6 7)",OIF)
elif x < 28:
nco = npcrroef[2][3]+npcrroef[2][x-21]+npcrroef[3][x-21]
OIF=nstd[x]/nco
print("OIF (3 4 ",x-20,OIF)
elif x < 30:
nco=npcrroef[2][4]+npcrroef[2][x-23]+npcrroef[4][x-23]
OIF=nstd[x]/nco
print("OIF (3 5 ",x-22," )",OIF)
elif x < 31:
nco=npcrroef[2][5]+npcrroef[2][6]+npcrroef[5][6]
OIF=nstd[x]/nco
print("OIF (3 6 7)",OIF)
elif x < 33:
nco=npcrroef[3][4]+npcrroef[3][x-26]+npcrroef[4][x-26]
OIF=nstd[x]/nco
print("OIF (4 5 ",x-25," )",OIF)
elif x<34:
nco=npcrroef[3][5]+npcrroef[3][6]+npcrroef[5][6]
OIF=nstd[x]/nco
print("OIF (4 6 7)",OIF)
elif x <35:
nco=npcrroef[4][5]+npcrroef[4][6]+npcrroef[5][6]
OIF=nstd[x]/nco
print("OIF( 5 6 7)",OIF)
若要运行OIF,需先求得标准差 ! (std)
PS: 求7波段OIF代码,个人认为非常冗杂,也没想到其他办法 去求,如若有知道的好友,麻烦告知!!!