def __call__(self, batch):
nir_image_batch, rgb_image_batch, theta_batch, image_name = batch[
'nir_image'], batch['rgb_image'], batch['theta'], batch['name']
if self.use_cuda:
nir_image_batch = nir_image_batch.cuda()
rgb_image_batch = rgb_image_batch.cuda()
theta_batch = theta_batch.cuda()
b, c, h, w = nir_image_batch.size()
# 为较大的采样区域生成对称填充图像
rgb_image_batch_pad = symmetricImagePad(rgb_image_batch,
self.padding_factor,
use_cuda=self.use_cuda)
nir_image_batch_pad = symmetricImagePad(nir_image_batch,
self.padding_factor,
use_cuda=self.use_cuda)
# convert to variables 其中Tensor是原始数据，并不知道梯度计算等问题，
# Variable里面有data，grad和grad_fn，其中data就是Tensor
# image_batch = Variable(image_batch, requires_grad=False)
# theta_batch = Variable(theta_batch, requires_grad=False)
# indices_R = torch.tensor([choice(self.channel_choicelist)])
# indices_G = torch.tensor([choice(self.channel_choicelist)])
# indices_R = torch.tensor([1])
# indices_G = torch.tensor([0])
#
# if self.use_cuda:
# indices_R = indices_R.cuda()
# indices_G = indices_G.cuda()
#
# rgb_image_batch_pad = torch.index_select(rgb_image_batch_pad, 1, indices_R)
# nir_image_batch_pad = torch.index_select(nir_image_batch_pad, 1, indices_G)
# 获取裁剪的图像
cropped_image_batch = self.rescalingTnf(
rgb_image_batch_pad, None, self.padding_factor,
self.crop_factor) # Identity is used as no theta given
# 获取裁剪变换的图像
warped_image_batch = self.geometricTnf(
nir_image_batch_pad, theta_batch, self.padding_factor,
self.crop_factor) # Identity is used as no theta given
return {
'source_image': cropped_image_batch,
'target_image': warped_image_batch,
'theta_GT': theta_batch,
'name': image_name
}
def save_matlab_pic(image_data, theta_aff):
image_batch = torch.from_numpy(image_data).transpose(1, 2).transpose(
0, 1).unsqueeze(1).float()
vis.showImageBatch(image_batch,
win='image_batch',
title='raw_image_batch',
start_index=16)
crop_factor = 9 / 16
padding_factor = 0.6
padding_image_batch = symmetricImagePad(image_batch,
padding_factor=padding_factor)
affTnf = AffineTnf(240, 240, use_cuda=False)
# 变换以后超出范围自动变为0
source_image_batch = affTnf(padding_image_batch, None, padding_factor,
crop_factor)
target_image_batch = affTnf(padding_image_batch, theta_aff, padding_factor,
crop_factor)
vis.showImageBatch(source_image_batch,
win='source_image_batch',
title='source_image_batch',
start_index=16)
vis.showImageBatch(target_image_batch,
win='target_image_batch',
title='target_image_batch',
start_index=16)
save_image_tensor(source_image_batch[16], 'mul_1s_s.png')
save_image_tensor(target_image_batch[16], 'mul_1t_s.png')
def get_image_information(image_dir, image_name, label_path, vis):
image_path = os.path.join(image_dir, image_name)
image_np = io.imread(image_path)
csv_data = read_csv_file(label_path)
label_row_param = csv_data.loc[csv_data['image'] == image_name].values
label_row_param = np.squeeze(label_row_param)
if image_name != label_row_param[0]:
raise ValueError("图片文件名和label图片文件名不匹配")
theta_aff = torch.from_numpy(label_row_param[1:].reshape(2, 3).astype(
np.float32)).unsqueeze(0)
image_batch = torch.from_numpy(image_np).transpose(1, 2).transpose(
0, 1).unsqueeze(0).float()
vis.showImageBatch(image_batch, win='image_batch', title='raw_image_batch')
crop_factor = 9 / 16
padding_factor = 0.6
# crop_factor = 3
# padding_factor = 0.9
padding_image_batch = symmetricImagePad(image_batch,
padding_factor=padding_factor)
affTnf = AffineTnf(446, 640, use_cuda=False)
# 变换以后超出范围自动变为0
source_image_batch = affTnf(padding_image_batch, theta_aff, padding_factor,
crop_factor)
target_image_batch = affTnf(padding_image_batch, None, padding_factor,
crop_factor)
# inverse_theta_aff = inverse_theta(theta_aff,use_cuda=False)
warped_image_batch = affTnf(target_image_batch,
theta_aff,
crop_factor=1,
padding_factor=1)
vis.showImageBatch(source_image_batch,
win='source_image_batch',
title='source_image_batch')
vis.showImageBatch(target_image_batch,
win='target_image_batch',
title='target_image_batch')
vis.showImageBatch(warped_image_batch,
win='warped_image_batch',
title='warped_image_batch')
# save_image_tensor(image_batch,'raw.jpg')
save_image_tensor(source_image_batch, 'source.jpg')
save_image_tensor(target_image_batch, 'target.jpg')
save_image_tensor(warped_image_batch, 'warped2.jpg')
def __call__(self, batch):
image_batch, theta_batch, image_name = batch['image'], batch[
'theta'], batch['name']
indices_R = torch.tensor([0])
indices_G = torch.tensor([2])
if self.use_cuda:
image_batch = image_batch.cuda()
theta_batch = theta_batch.cuda()
indices_R = indices_R.cuda()
indices_G = indices_G.cuda()
# 对图像边缘进行镜像填充
image_batch = symmetricImagePad(image_batch,
self.padding_factor,
use_cuda=self.use_cuda)
# 获得单通道图片，沿着指定维度对输入进行切片，取index中指定的相应项(index为一个LongTensor)，
# 然后返回到一个新的张量， 返回的张量与原始张量_Tensor_有相同的维度(在指定轴上)。
image_batch_R = torch.index_select(image_batch, 1, indices_R)
image_batch_G = torch.index_select(image_batch, 1, indices_G)
# 原始图像R通道缩放
original_image_batch = self.affineTnf(image_batch_R, None,
self.padding_factor,
self.crop_factor)
wraped_image_batch = self.affineTnf(image_batch_G, theta_batch,
self.padding_factor,
self.crop_factor)
pair_result = {
'source_image': original_image_batch,
'target_image': wraped_image_batch,
'theta_GT': theta_batch,
'name': image_name
}
return pair_result
def __call__(self, batch):
image_batch, theta_batch,image_name = batch['image'], batch['theta'],batch['name']
if self.use_cuda:
image_batch = image_batch.cuda()
theta_batch = theta_batch.cuda()
b, c, h, w = image_batch.size()
# 为较大的采样区域生成对称填充图像
image_batch = symmetricImagePad(image_batch, self.padding_factor,use_cuda=self.use_cuda)
# indices_R = torch.tensor([choice(self.channel_choicelist)])
# indices_G = torch.tensor([choice(self.channel_choicelist)])
indices_R = torch.tensor([0])
indices_G = torch.tensor([2])
#
if self.use_cuda:
indices_R = indices_R.cuda()
indices_G = indices_G.cuda()
image_batch_R = torch.index_select(image_batch, 1, indices_R)
image_batch_G = torch.index_select(image_batch, 1, indices_G)
image_batch_R = torch.cat((image_batch_R,image_batch_R,image_batch_R),1)
image_batch_G = torch.cat((image_batch_G,image_batch_G,image_batch_G),1)
# 获取裁剪的图像
cropped_image_batch = self.rescalingTnf(image_batch_R, None, self.padding_factor,
self.crop_factor) # Identity is used as no theta given
# 获取裁剪变换的图像
warped_image_batch = self.geometricTnf(image_batch_G, theta_batch,
self.padding_factor,
self.crop_factor) # Identity is used as no theta given
return {'source_image': cropped_image_batch, 'target_image': warped_image_batch, 'theta_GT': theta_batch,
'name':image_name}