def test():
test_path = os.path.join(os.getcwd(), TEST_IMAGE_FOLDER)
imgs_path_array = glob.glob(test_path + "/*.jpg")
psnr_arr = []
for img_path in imgs_path_array:
lr_img_data, hr_img_data = preprocessing.process_original(img_path)
sr_img_data = rebuild(lr_img_data)
psnr_arr.append(PSNR.cal_psnr(hr_img_data, sr_img_data))
"""
lr_path = img_path.split('.')[0] + "_LR." + img_path.split('.')[-1]
sr_path = img_path.split('.')[0] + "_SR." + img_path.split('.')[-1]
lr_img_rgb_data = preprocessing.ycbcr2rgb(lr_img_data)
sr_img_rgb_data = preprocessing.ycbcr2rgb(sr_img_data)
lr_img_rgb = Image.fromarray(lr_img_rgb_data.astype('uint8'))
sr_img_rgb = Image.fromarray(sr_img_rgb_data.astype('uint8'))
lr_img_rgb.save(lr_path)
sr_img_rgb.save(sr_path)
"""
psnr_mean = np.sum(psnr_arr) / len(psnr_arr)
#print(psnr_arr)
print("This model's PSNR: [%.4f]" % (psnr_mean))
with open("PSNR.txt", 'a') as file:
file.write("---------------------------------------------------\n")
file.write(time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) + '\n')
file.write("Ratio: " + str(preprocessing.RATIO) + "\n")
file.write("Activation Function: relu\n")
file.write("Epoch: " + espcn.get_epoch() + "\tBatch_size: " + espcn.get_batch_size() + "\n")
file.write("Test images set: " + TEST_IMAGE_FOLDER .split('/')[-1] + "\n")
file.write("Average PSNR: [%.4f]\n" % (psnr_mean))
file.write("---------------------------------------------------\n\n\n")
def main():
img = cv2.imread('../data/Test/Set14/baboon.bmp')
img = cv2.cvtColor(img, cv2.COLOR_BGR2YUV)[:, :, 0]
m, n = img.shape
# convert to matlab ycbcr
# for i in range(m):
# for j in range(n):
# img[i,j] = 16 + 219.*(img[i,j]/255.)
img_lr = bicubic(img, int(m / 3), int(n / 3))
img_hr = bicubic(img_lr, m, n)
# img_cv = cv2.resize(img, (int(n/3), int(m/3)), interpolation=cv2.INTER_CUBIC)
# img_cv = cv2.resize(img_cv, (n, m), interpolation=cv2.INTER_CUBIC)
# cv2.imshow('bicubic', img_hr)
# cv2.imshow('cv2', img_cv)
# cv2.waitKey()
print PSNR.PSNR(img, img_hr)
def main():
args = get_args()
Atest, Btest = data.train_dataset(args.dir, args.batch_size,
args.image_size, 1)
B_test_iter = iter(Btest)
A_test_iter = iter(Atest)
B_test = Variable(B_test_iter.next()[0])
A_test = Variable(A_test_iter.next()[0])
G_12 = model.Generator(64)
G_21 = model.Generator(64)
checkpoint = torch.load(args.state_dict)
G_12.load_state_dict(checkpoint['G_12_state_dict'])
G_21.load_state_dict(checkpoint['G_21_state_dict'])
if torch.cuda.is_available():
test = test.cuda()
noised = noised.cuda()
G_12 = G_12.cuda()
G_21 = G_21.cuda()
G_12.eval()
G_21.eval()
generate_A_image = G_21(B_test.float())
grid = vutils.make_grid(generate_A_image, nrow=8, normalize=True)
vutils.save_image(grid, "generate_A_image.png")
generate_B_image = G_12(A_test.float())
grid = vutils.make_grid(generate_A_image, nrow=8, normalize=True)
vutils.save_image(grid, "generate_B_image.png")
loss = PSNR.PSNR()
estimate_loss_generate_A = loss(generate_A_image, A_test)
estimate_loss_generate_B = loss(generate_B_image, B_test)
print(estimate_loss_generate_A)
print(estimate_loss_generate_B)
def main():
args = get_args()
Atest, Btest = data.train_dataset(args.dir, args.batch_size, args.image_size, 1000)
B_test_iter = iter(Btest)
A_test_iter = iter(Atest)
G_21 = model.Generator(args.batch_size)
#G_21 = residual_model.Generator(1,1)
checkpoint = torch.load(args.state_dict)
G_21.load_state_dict(checkpoint['G_21_state_dict'])
estimate_loss_generate = 0
for i in range(1000):
B_test = Variable(B_test_iter.next()[0])
A_test = Variable(A_test_iter.next()[0])
grid = vutils.make_grid(B_test, nrow=8)
vutils.save_image(grid,"B_image.png")
if torch.cuda.is_available():
B_test = B_test.cuda()
A_test = A_test.cuda()
G_21 = G_21.cuda()
G_21.eval()
generate_A_image = G_21(B_test.float())
grid = vutils.make_grid(generate_A_image, nrow=8)
vutils.save_image(grid,"generate_A_image.png")
loss = PSNR.PSNR()
estimate_loss_generate = estimate_loss_generate +loss(generate_A_image, A_test)
estimate_loss_generate = estimate_loss_generate /1000
print(estimate_loss_generate)
def do_filtering(img_name, folder, file):
img = cv2.imread(img_name, cv2.IMREAD_GRAYSCALE)
folder = folder.replace('whole_raw_image', 'bm3d')
e1 = cv2.getTickCount() # cv2.getTickCount
Basic_img = BM3D_1st_step(img)
e2 = cv2.getTickCount()
time = (e2 - e1) / cv2.getTickFrequency()
print("The Processing time of the First step is %f s" % time)
Final_img = BM3D_2nd_step(Basic_img, img)
e3 = cv2.getTickCount()
time = (e3 - e2) / cv2.getTickFrequency()
print("The Processing time of the Second step is %f s" % time)
filename = folder + file
print(filename)
cv2.imwrite(filename, Final_img)
psnr = PSNR.PSNR(img, Final_img)
print("The PSNR between the two img of the Second step is %f" % psnr)
time = (e3 - e1) / cv2.getTickFrequency()
print("The total Processing time is %f s" % time)
def calALL():
path = opt.folders
ps = []
ss = []
list_PSNR =[]
list_SSIM =[]
file_list = os.listdir(path)
f = open(path+"average.txt", 'w')
f.write('average (PSNR,SSIM)\n')
f.close()
file_list = sorted(file_list, key = len)
for Npath in file_list:
newPath = path+Npath+'/'
averPSNR = 0
averSSIM = 0
count = 0
if not (search(newPath)):
continue
else:
f = open(newPath+"PSNR.txt", 'w')
f.close()
f = open(newPath+"SSIM.txt", 'w')
f.close()
print("*folder* = "+newPath)
newfile_list = os.listdir(newPath)
list_PSNR =[]
list_SSIM =[]
for i in newfile_list:
if(i[-9:]=='epoch.jpg'):
os.remove(newPath+'epoch.jpg')
continue
if(i[-3:]=='txt' or i[-8:]=='SSIM.jpg' or i[-8:]=='PSNR.jpg'):
continue
img = i
LR = newPath + img
print('open '+ LR)
HR = opt.HR
HR = HR + img
print('open '+HR)
#biimg = cv2.imread(LR)
#biimg = cv2.resize(biimg, None, fx=4, fy=4, interpolation=cv2.INTER_CUBIC)
#cv2.imwrite(opt.HR+'../bicubic/'+i ,biimg)
#LR = opt.HR+'../bicubic/'+i
#print(LR)
try:
one, two = PSNR.cal_PSNRandSSIM(HR, LR)
except:
print('skip')
continue
print(f"{one} is PSNR, {two} is SSIM")
f = open(newPath+"PSNR.txt", 'a')
f.write(img+f" PSNR is {one}\n")
f.close()
f = open(newPath+"SSIM.txt", 'a')
f.write(img+f" SSIM is {two}\n")
f.close()
list_PSNR.append(one)
list_SSIM.append(two)
#well.. im not good at python..
averPSNR += one
averSSIM += two
count +=1
print('=============================')
graphing.graph(list_PSNR, [], newPath,'PSNR',opt.picofname)
graphing.graph(list_SSIM, [], newPath,'SSIM',opt.picofname)
if(count==0):
averPSNR = 0
averSSIM = 0
else:
averPSNR /= count
averSSIM /= count
print("{} is average PSNR, {} is average SSIM".format(averPSNR,averSSIM))
ps.append(averPSNR)
ss.append(averSSIM)
graphing.graph(ps, [], path,'PSNR',opt.X)
graphing.graph(ss, [], path,'SSIM',opt.X)
f = open(path+"average.txt", 'a')
f.write("{} : {} is average PSNR, {} is average SSIM\n".format(Npath,averPSNR,averSSIM))
f.close()
if __name__ == '__main__':
cv2.setUseOptimized(True) # OpenCV 中的很多函数都被优化过（使用 SSE2，AVX 等）。也包含一些没有被优化的代码。使用函数 cv2.setUseOptimized() 来开启优化。
img_name = "C:/Users/admin/Desktop/BM3D-Denoise/BM3D_test_images/lw3.png" # 图像的路径
img = cv2.imread(img_name, cv2.IMREAD_GRAYSCALE) # 读入图像，cv2.IMREAD_GRAYSCALE:以灰度模式读入图像
# 记录程序运行时间
e1 = cv2.getTickCount() # cv2.getTickCount 函数返回从参考点到这个函数被执行的时钟数
# if(img is not None):
# print("success")
Basic_img = BM3D_1st_step(img)
e2 = cv2.getTickCount()
time = (e2 - e1) / cv2.getTickFrequency() # 计算函数执行时间
print ("The Processing time of the First step is %f s" % time)
cv2.imwrite("Basic3.jpg", Basic_img)
psnr = PSNR.PSNR(img, Basic_img)
print ("The PSNR between the two img of the First step is %f" % psnr)
# Basic_img = cv2.imread("Basic3.jpg", cv2.IMREAD_GRAYSCALE)
Final_img = BM3D_2nd_step(Basic_img, img)
e3 = cv2.getTickCount()
time = (e3 - e2) / cv2.getTickFrequency()
print ("The Processing time of the Second step is %f s" % time)
cv2.imwrite("Final3.jpg", Final_img)
psnr = PSNR.PSNR(img, Final_img)
print ("The PSNR between the two img of the Second step is %f" % psnr)
time = (e3 - e1) / cv2.getTickFrequency()
print ("The total Processing time is %f s" % time)
"./data/test_set_s3.h5") #h5数据集制作工具在data中，可自己制作
test_loader = Data.DataLoader(dataset=test_set,
num_workers=1,
batch_size=40,
shuffle=False)
net = torch.load("./checkpoint/model_epoch_100.pth")["model"]
net.cpu()
i = random.randint(0, 100)
aa, bb = test_loader.dataset[i]
lim1 = tran_im(aa)
label = tran_im(bb)
lim1.show()
print("CUBIC_PSNR:", PSNR.psnr(lim1, label))
T_X = torch.unsqueeze(tran_ten(lim1), 0)
prediction = net(T_X)
prediction[prediction < 0] = 0
prediction[prediction > 1] = 1
lim2 = tran_im(prediction[0])
lim2.show()
print("NET_PSNR:", PSNR.psnr(lim2, label))
label.show()
data = MSE_hybnetd[j]
dd = np.ravel(data)
ivmin = min(ivmin, np.amin(dd))
ivmax = max(ivmax, np.amax(dd))
images.append(a.imshow(data, cmap=cm.viridis))
if j <= 4:
a.add_patch(plt.Rectangle((curve1_pos_x[j] - 15, curve1_pos_y[j] - 15), width=31, height=31, linewidth=1,
edgecolor='r', facecolor='none'))
a.add_patch(plt.Rectangle((curve2_pos_x[j] - 15, curve2_pos_y[j] - 15), width=31, height=31, linewidth=1,
edgecolor='lime', facecolor='none'))
else: # Coordinates of cutted ICVL images has 15 pixels right shift
a.add_patch(plt.Rectangle((curve1_pos_x[j] - 15 - 15, curve1_pos_y[j] - 15), width=31, height=31, linewidth=1,
edgecolor='r', facecolor='none'))
a.add_patch(plt.Rectangle((curve2_pos_x[j] - 15 - 15, curve2_pos_y[j] - 15), width=31, height=31, linewidth=1,
edgecolor='lime', facecolor='none'))
a.text(40, 70, 'PSNR=' + str('%.2f' % PSNR.mse2psnr(np.mean(dd), bitdepth=1)) + 'dB',
fontproperties=FontProperties(size=5), color='w')
if i == 2:
a.axis('off')
data = MSE_hsnetd[j]
dd = np.ravel(data)
ivmin = min(ivmin, np.amin(dd))
ivmax = max(ivmax, np.amax(dd))
images.append(a.imshow(data, cmap=cm.viridis))
if j <= 4:
a.add_patch(plt.Rectangle((curve1_pos_x[j] - 15, curve1_pos_y[j] - 15), width=31, height=31, linewidth=1,
edgecolor='r', facecolor='none'))
a.add_patch(plt.Rectangle((curve2_pos_x[j] - 15, curve2_pos_y[j] - 15), width=31, height=31, linewidth=1,
edgecolor='lime', facecolor='none'))
else: # Coordinates of cutted ICVL images has 15 pixels right shift
a.add_patch(plt.Rectangle((curve1_pos_x[j] - 15 - 15, curve1_pos_y[j] - 15), width=31, height=31, linewidth=1,