【深度学习笔记】理解Bicubic，双三次插值

最新推荐文章于 2022-11-27 21:50:25 发布

月满星沉

最新推荐文章于 2022-11-27 21:50:25 发布

阅读量1k

点赞数

分类专栏：深度学习 Python 学习学习笔记文章标签： Bicubic 超分辨率

本文链接：https://blog.csdn.net/Hunter_Murphy/article/details/106597649

版权

学习笔记同时被 3 个专栏收录

84 篇文章 8 订阅

订阅专栏

深度学习

47 篇文章 4 订阅

订阅专栏

Python 学习

23 篇文章 0 订阅

订阅专栏

一、源码1

# -*- coding: utf-8 -*-
# @Time    : 2020/10/12 22:45
# @Author  : Murphy
# @Email   : Murphy.AI@foxmail.com
# from https://github.com/yiyang7/Super_Resolution_with_CNNs_and_GANs/tree/master/model

import torch.nn as nn

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()

    def forward(self, x):
        out = nn.functional.upsample(x, size=None, scale_factor=2, mode='nearest')

        return out

二、源码2

from PIL import Image
import os

def Bicubic():
    path = r"source path"
    save_path = r"target path of the results"
    filelist = os.listdir(path)
    for files in filelist:
        img_path = os.path.join(path, files)
        img = Image.open(img_path)
        DealImage = img.resize((int(img.size[0] * 4), int(img.size[1] * 4)), Image.BICUBIC) # change the scale factor here
        files = files.split('.')[0] + '.png'
        img_path = os.path.join(save_path, files)
        DealImage.save(img_path)

if __name__ == '__main__':
    Bicubic()