utils.py

#!/usr/bin/env python
# -*- coding:utf-8 -*-
# Power by Zongsheng Yue 2019-01-22 22:07:08

import torch
import torch.nn as nn
from torch.autograd import Function as autoF
from scipy.special import gammaln
from skimage.measure import compare_psnr, compare_ssim
from skimage import img_as_ubyte
import numpy as np
import sys
from math import floor

def ssim_index(im1, im2):
    '''
    Input:
        im1, im2: np.uint8 format
    '''
    if im1.ndim == 2:
        out = compare_ssim(im1, im2, data_range=255, gaussian_weights=True,
                                                    use_sample_covariance=False, multichannel=False)
    elif im1.ndim == 3:
        out = compare_ssim(im1, im2, data_range=255, gaussian_weights=True,
                                                     use_sample_covariance=False, multichannel=True)
    else:
        sys.exit('Please input the corrected images')
    return out

def im2patch(im, pch_size, stride=1):
    '''
    Transform image to patches.
    Input:
        im: 3 x H x W or 1 X H x W image, numpy format
        pch_size: (int, int) tuple or integer
        stride: (int, int) tuple or integer
    '''
    if isinstance(pch_size, tuple):
        pch_H, pch_W = pch_size
    elif isinstance(pch_size, int):
        pch_H = pch_W = pch_size
    else:
        sys.exit('The input of pch_size must be a integer or a int tuple!')

    if isinstance(stride, tuple):
        stride_H, stride_W = stride
    elif isinstance(stride, int):
        stride_H = stride_W = stride
    else:
        sys.exit('The input of stride must be a integer or a int tuple!')

    C, H, W = im.shape
    num_H = len(range(0, H-pch_H+1, stride_H))
    num_W = len(range(0, W-pch_W+1, stride_W))
    num_pch = num_H * num_W
    pch = np.zeros((C, pch_H*pch_W, num_pch), dtype=im.dtype)
    kk = 0
    for ii in range(pch_H):
        for jj in range(pch_W):
            temp = im[:, ii:H-pch_H+ii+1:stride_H, jj:W-pch_W+jj+1:stride_W]
            pch[:, kk, :] = temp.reshape((C, num_pch))
            kk += 1

    return pch.reshape((C, pch_H, pch_W, num_pch))

def batch_PSNR(img, imclean):
    Img = img.data.cpu().numpy()
    Iclean = imclean.data.cpu().numpy()
    Img = img_as_ubyte(Img)
    Iclean = img_as_ubyte(Iclean)
    PSNR = 0
    for i in range(Img.shape[0]):
        PSNR += compare_psnr(Iclean[i,:,:,:], Img[i,:,:,:], data_range=255)
    return (PSNR/Img.shape[0])

def batch_SSIM(img, imclean):
    Img = img.data.cpu().numpy()
    Iclean = imclean.data.cpu().numpy()
    Img = img_as_ubyte(Img)
    Iclean = img_as_ubyte(Iclean)
    SSIM = 0
    for i in range(Img.shape[0]):
        SSIM += ssim_index(Iclean[i,:,:,:].transpose((1,2,0)), Img[i,:,:,:].transpose((1,2,0)))
    return (SSIM/Img.shape[0])

def peaks(n):
    '''
    Implementation the peak function of matlab.
    '''
    X = np.linspace(-3, 3, n)
    Y = np.linspace(-3, 3, n)
    [XX, YY] = np.meshgrid(X, Y)
    ZZ = 3 * (1-XX)**2 * np.exp(-XX**2 - (YY+1)**2) \
            - 10 * (XX/5.0 - XX**3 -YY**5) * np.exp(-XX**2-YY**2) - 1/3.0 * np.exp(-(XX+1)**2 - YY**2)
    return ZZ

def generate_gauss_kernel_mix(H, W):
    '''
    Generate a H x W mixture Gaussian kernel with mean (center) and std (scale).
    Input:
        H, W: interger
        center: mean value of x axis and y axis
        scale: float value
    '''
    pch_size = 32
    K_H = floor(H / pch_size)
    K_W = floor(W / pch_size)
    K = K_H * K_W
    # prob = np.random.dirichlet(np.ones((K,)), size=1).reshape((1,1,K))
    centerW = np.random.uniform(low=0, high=pch_size, size=(K_H, K_W))
    ind_W = np.arange(K_W) * pch_size
    centerW += ind_W.reshape((1, -1))
    centerW = centerW.reshape((1,1,K)).astype(np.float32)
    centerH = np.random.uniform(low=0, high=pch_size, size=(K_H, K_W))
    ind_H = np.arange(K_H) * pch_size
    centerH += ind_H.reshape((-1, 1))
    centerH = centerH.reshape((1,1,K)).astype(np.float32)
    scale = np.random.uniform(low=pch_size/2, high=pch_size, size=(1,1,K))
    scale = scale.astype(np.float32)
    XX, YY = np.meshgrid(np.arange(0, W), np.arange(0,H))
    XX = XX[:, :, np.newaxis].astype(np.float32)
    YY = YY[:, :, np.newaxis].astype(np.float32)
    ZZ = 1./(2*np.pi*scale**2) * np.exp( (-(XX-centerW)**2-(YY-centerH)**2)/(2*scale**2) )
    out = ZZ.sum(axis=2, keepdims=False) / K

    return out

def sincos_kernel():
    # Nips Version
    [xx, yy] = np.meshgrid(np.linspace(1, 10, 256), np.linspace(1, 20, 256))
    # [xx, yy] = np.meshgrid(np.linspace(1, 10, 256), np.linspace(-10, 15, 256))
    zz = np.sin(xx) + np.cos(yy)
    return zz

def capacity_cal(net):
    out = 0
    for param in net.parameters():
        out += param.numel()*4/1024/1024
    # print('Networks Parameters: {:.2f}M'.format(out))
    return out

class LogGamma(autoF):
    '''
    Implement of the logarithm of gamma Function.
    '''
    @staticmethod
    def forward(ctx, input):
        ctx.save_for_backward(input)
        if input.is_cuda:
            input_np = input.detach().cpu().numpy()
        else:
            input_np = input.detach().numpy()
        out = gammaln(input_np)
        out = torch.from_numpy(out).to(device=input.device).type(dtype=input.dtype)

        return out

    @staticmethod
    def backward(ctx, grad_output):
        input, = ctx.saved_tensors
        grad_input = torch.digamma(input) * grad_output

        return grad_input

def load_state_dict_cpu(net, state_dict0):
    state_dict1 = net.state_dict()
    for name, value in state_dict1.items():
        assert 'module.'+name in state_dict0
        state_dict1[name] = state_dict0['module.'+name]
    net.load_state_dict(state_dict1)