!pip install spectral

import matplotlib
matplotlib.rcParams['figure.figsize'] = (15,10)

%config InlineBackend.figure_format = 'retina'


!pip install spectral

# Bibliotecas
import tifffile as tif
import matplotlib.pyplot as plt
from spectral import imshow
import numpy as np

# Lendo imagem
img = tif.imread('Cubo_SR.tif')

# Visualizando composição colorida
imshow(img, (2,1,0), stretch=(0.02), figsize=(15,10))

# Definindo espectros
mata = img[600,1400,:]
cerrado = img[600,600,:]
urbano = img[1000,200,:]
agua = img[500,700,:]

espectros = [mata, cerrado, urbano, agua]

# Visualizando epectros
bandas = ['Blue', 'Green', 'Red', 'NIR', 'SWIR-1','SWIR-2']
plt.plot(bandas,img[600,1400,:])

# Visualizando espectros separados
cores = ['Darkgreen', 'lime', 'Red', 'Blue']
nomes = ['Mata', 'Cerrado', 'Urbano', 'Água']
fig, ax = plt.subplots(4, 1, sharex=True, figsize=(15,10))

for i in range(len(espectros)):
  ax[i].plot(bandas, espectros[i], color=cores[i])
  ax[i].set_title(nomes[i])

# Visualizando espectros em uma única figura
fig2, ax2 = plt.subplots()

for i in range(len(espectros)):
  ax2.plot(bandas, espectros[i], color=cores[i])
  ax2.legend(nomes, loc="lower left")

# Calculando NDVI
ndvi = ( img[:,:,3] - img[:,:,2] )  /  ( img[:,:,3] + img[:,:,2] )

# Visualizando de forma contínua
plt.imshow(ndvi, cmap='RdYlGn')
plt.colorbar()

from matplotlib import colors

# Visualizando de forma discreta

cmap = colors.ListedColormap(['red', 'yellow','orange','lime','darkgreen'])

norm = colors.BoundaryNorm(bounds, cmap.N)

bounds=[np.min(ndvi),0.03,0.25,0.5,0.75,np.max(ndvi)]

norm = colors.BoundaryNorm(bounds, cmap.N)

im = plt.imshow(ndvi,cmap=cmap, norm=norm)
plt.colorbar(im)

# Reclassificando ndvi
ndvi_reclass= ndvi.copy()


ndvi_reclass[ndvi <= 0.03] = 1
ndvi_reclass[(ndvi > 0.03) & (ndvi <= 0.25)] = 2
ndvi_reclass[(ndvi > 0.25) & (ndvi <= 0.50)] = 3
ndvi_reclass[(ndvi > 0.50) & (ndvi <= 0.75)] = 4
ndvi_reclass[(ndvi > 0.75)] = 5

plt.imshow(ndvi_reclass, cmap='RdYlGn')