使3D高斯拟合数据集

Question

I have a data file with first column x, second coulmn y and third column z. I can call these values via

x=mat0[:,0]

That is not the problem. I can also create and plot a 3D Gaussian with these data or (as you see in my script below) via definition of the function "twoD_Gauss".

Now I want to fit this function "twoD_Gauss" to the dataset (x,y,z) and print out the values for amplitude sigma etc.

This is what I got:

from matplotlib import pyplot;
from pylab import genfromtxt;  
import matplotlib.pyplot as plt
import numpy as np
from mpl_toolkits.mplot3d import Axes3D
from numpy.random import randn
from scipy import array, newaxis

# Load file into mat0
mat0 = genfromtxt("0005.map");
fig = plt.figure(figsize=(20,10))

############ 3D  ###############
ax = fig.add_subplot(1, 2, 2, projection='3d')
#Load data
mat0 = genfromtxt("0005.map");

# define Gaussian
def twoD_Gauss((x,y),amplitude,x0,y0,sigma_x,sigma_y,offset):
    x0=float(x0)
    y0=float(y0)
    return offset + amplitude*np.exp(-(((x-x0)**(2)/(2*sigma_x**(2))) + ((y-y0)**(2)/(2*sigma_y**(2)))))

#define x and y and z (z not used, x and y shifted)
x = mat0[:,0]-150
y = mat0[:,1]-143
z = mat0[:,2]
#create data
data = twoD_Gauss((x, y), 15, 0, 0, 20, 20, 10)

# plot twoD_Gaussian data generated above
ax = plt.axes(projection='3d')
ax.plot_trisurf(x, y, data, cmap="jet", linewidth=0)

#FITTING HELP!
initial_guess = (24000,0,0,25,25,6000)
params, pcov = opt.curve_fit(twoD_Gauss, (x,y), data,initial_guess)
print(params)

plt.show()

I think I did it correct, but its actually not fitting.

The printed params are the parameters I gave in data.

Answer 1

Ok I found the solution myself:

My problem was that I fitted to the data but data was already defined. So what I did is, I changed data to the z-data given in my file.

data=mat0[:,2]

Now the curve_fit fits the twoD_Gauss via (x,y) to the given z-values.