可见光和红外图像的融合
3种较为简单的可见光和红外图像融合方式，包括固定权重的融合、TIF方式等：
import pywt
import cv2
import numpy as np
import threading
class ImgFusion:
def TIF_algo(self,p1, p2, median_blur_value=3, mean_blur_value=35):
p1_b = cv2.blur(p1, (mean_blur_value, mean_blur_value))
p1_b = p1_b.astype(np.float)
p2_b = cv2.blur(p2, (mean_blur_value, mean_blur_value))
p2_b = p2_b.astype(np.float)
# cv2.imshow('picture after mean blur p1_b', p1_b)
# cv2.imshow('picture after mean blur p2_b', p2_b)
# p1_d = abs(p1.astype(np.float) - p1_b)
# p2_d = abs(p2.astype(np.float) - p2_b)
p1_d = p1.astype(np.float) - p1_b
p2_d = p2.astype(np.float) - p2_b
# cv2.imshow('detail layer p1', p1_d / 255.0)
# cv2.imshow('detail layer p2', p2_d / 255.0)
p1_after_medianblur = cv2.medianBlur(p1, median_blur_value)
p2_after_medianblur = cv2.medianBlur(p2, median_blur_value)
# cv2.imshow('picture after median blur p1_after_medianblur', p1_after_medianblur)
# cv2.imshow('picture after median blur p2_after_medianblur', p2_after_medianblur)
p1_after_medianblur = p1_after_medianblur.astype(np.float)
p2_after_medianblur = p2_after_medianblur.astype(np.float)
p1_subtract_from_median_mean = p1_after_medianblur - p1_b + 0.0001
p2_subtract_from_median_mean = p2_after_medianblur - p2_b + 0.0001
# cv2.imshow('subtract_from_median_mean p1_subtract_from_median_mean', p1_subtract_from_median_mean/255.0)
# cv2.imshow('subtract_from_median_mean p2_subtract_from_median_mean', p2_subtract_from_median_mean/255.0)
m1 = p1_subtract_from_median_mean[:, :, 0]
m2 = p1_subtract_from_median_mean[:, :, 1]
m3 = p1_subtract_from_median_mean[:, :, 2]
res = m1 * m1 + m2 * m2 + m3 * m3
# delta1 = np.sqrt(res)
delta1 = res
m1 = p2_subtract_from_median_mean[:, :, 0]
m2 = p2_subtract_from_median_mean[:, :, 1]
m3 = p2_subtract_from_median_mean[:, :, 2]
res = m1 * m1 + m2 * m2 + m3 * m3
delta2 = abs(m1)
delta_total = delta1 + delta2
psi_1 = delta1 / delta_total
psi_2 = delta2 / delta_total
psi1 = np.zeros(p1.shape, dtype=np.float)
psi2 = np.zeros(p2.shape, dtype=np.float)
psi1[:, :, 0] = psi_1
psi1[:, :, 1] = psi_1
psi1[:, :, 2] = psi_1
psi2[:, :, 0] = psi_2
psi2[:, :, 1] = psi_2
psi2[:, :, 2] = psi_2
p_b = 0.5 * (p1_b + p2_b)
# cv2.imshow('base pic1', p1_b / 255.0)
# cv2.imshow('base pic2', p2_b / 255.0)
# cv2.imshow('base pic', p_b / 255.0)
p_d = psi1 * p1_d + psi2 * p2_d
# cv2.imshow('detail layer plus', p_d / 255.0)
# cv2.imshow('detail pic plus psi1 psi1 * p1_d', psi1 * p1_d)
# cv2.imshow('detail pic plus psi1 psi2 * p2_d', psi2 * p2_d)
p = p_b + p_d
# img = cv2.cvtColor(p, cv2.COLOR_BGR2RGB)
# cv2.imshow('final result', p / 255.0)
# cv2.imwrite('./final_res.jpg', p)
# cv2.waitKey(0)
return p
def strategy(self,arr_hvd1,arr_hvd2):
k1 = 0.8
k2 = 0.2
arr_w1 = np.where( np.abs(arr_hvd1) > np.abs(arr_hvd2), k1, k2)
arr_w2 = np.where( np.abs(arr_hvd1) < np.abs(arr_hvd2), k1, k2)
return arr_w1,arr_w2
def fusion(self, arr_visible, arr_infrared):
it_h1 = arr_visible.shape[0]
it_w1 = arr_visible.shape[1]
it_h2 = arr_infrared.shape[0]
it_w2 = arr_infrared.shape[1]
if it_h1 % 2 != 0:
it_h1 = it_h1 + 1
if it_w1 % 2 != 0:
it_w1 = it_w1 + 1
if it_h2 % 2 != 0:
it_h2 = it_h2 + 1
if it_w2 % 2 != 0:
it_w2 = it_w2 + 1
arr_visible = cv2.resize(arr_visible, (it_w1, it_h1))
arr_infrared = cv2.resize(arr_infrared, (it_w2, it_h2))
it_level = 5
arr_Gray1,arr_Gray2 = cv2.cvtColor(arr_visible, cv2.COLOR_BGR2GRAY), cv2.cvtColor(arr_infrared, cv2.COLOR_BGR2GRAY)
arr_Gray1 = arr_Gray1*1.0
arr_Gray2 = arr_Gray2*1.0
arr_visible = arr_visible*1.0
arr_infrared = arr_infrared*1.0
arr_decGray1 = pywt.wavedec2(arr_Gray1, 'sym4', level=it_level)
arr_decGray2 = pywt.wavedec2(arr_Gray2, 'sym4', level=it_level)
ls_decRed1 = pywt.wavedec2(arr_visible[:, :, 0], 'sym4', level=it_level)
ls_decGreen1 = pywt.wavedec2(arr_visible[:, :, 1], 'sym4', level=it_level)
ls_decBlue1 = pywt.wavedec2(arr_visible[:, :, 2], 'sym4', level=it_level)
ls_recRed = []
ls_recGreen = []
ls_recBlue = []
for it_i,(arr_gray1,arr_gray2,arr_red1,arr_green1,arr_blue1) in enumerate(zip(arr_decGray1,arr_decGray2,ls_decRed1,ls_decGreen1,ls_decBlue1)):
if it_i == 0:
fl_w1 = 0.5
fl_w2 = 0.5
us_recRed = fl_w1 * arr_red1 + fl_w2 * arr_gray2
us_recGreen = fl_w1 * arr_green1 + fl_w2 * arr_gray2
us_recBlue = fl_w1 * arr_blue1 + fl_w2 * arr_gray2
else:
us_recRed = []
us_recGreen = []
us_recBlue = []
for arr_grayHVD1,arr_grayHVD2,arr_redHVD1,arr_greenHVD1,arr_blueHVD1, in zip(arr_gray1,arr_gray2,arr_red1,arr_green1,arr_blue1):
arr_w1,arr_w2 = self.strategy(arr_grayHVD1,arr_grayHVD2)
arr_recRed = arr_w1*arr_redHVD1 + arr_w2*arr_grayHVD2
arr_recGreen = arr_w1*arr_greenHVD1 + arr_w2*arr_grayHVD2
arr_recBlue = arr_w1*arr_blueHVD1 + arr_w2*arr_grayHVD2
us_recRed.append(arr_recRed)
us_recGreen.append(arr_recGreen)
us_recBlue.append(arr_recBlue)
ls_recRed.append(us_recRed)
ls_recGreen.append(us_recGreen)
ls_recBlue.append(us_recBlue)
arr_rec = np.zeros(arr_visible.shape)
arr_rec[:, :, 0] = pywt.waverec2(ls_recRed, 'sym4')
arr_rec[:, :, 1] = pywt.waverec2(ls_recGreen, 'sym4')
arr_rec[:, :, 2] = pywt.waverec2(ls_recBlue, 'sym4')
return arr_rec
img_rgb=cv2.imread('./rgb.jpg')
img_t=cv2.imread('./t.jpg')
f=ImgFusion()
#five ways of fusion
img_f = f.TIF_algo(img_rgb, img_t)
#img_f = f.fusion(img_rgb,img_t)
#img_f =cv2.addWeighted(img_rgb,0.5, img_t,0.5,0)
#img_f = np.where(img_f > 255.0, 255.0,img_f)
#img_f = np.where(img_f < 0.0, 0.0,img_f)采用均分权重（代码里倒数第三行的）的融合效果：（可见光、红外、融合）
