先是看到 英雄联盟大拼图的文章,里面用爬虫爬了LOL官网所有英雄的皮肤图片,然后做了一个拼图,他的拼图用其他软件做的,然后我就想到python应该也能做拼图吧,然后就搜到 使用Python简单实现马赛克拼图这篇技术文,照猫画虎实践了一番。
爬虫
因为目前课题主要做数据清洗、可视化,对爬虫没什么要求,就暂时没深入学习,这段代码也没彻底看懂。。。233,因此只对部分代码做笔记整理
import requests
import json
import time
import re
headers = {
'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36'
}
def get_hero(url):
response = requests.get(url=url, headers=headers)
html = response.text
result = re.findall('if.*?champion=(.*?)data', html, re.S)
data = json.loads(result[0][:-2] + '}')
for key, value in data['keys'].items():
hero = value
get_skin(hero)
time.sleep(2) #推迟执行的秒数,应该是为了防止被反爬
def get_skin(hero):
url = 'http://lol.qq.com/biz/hero/' + hero + '.js'
response = requests.get(url=url, headers=headers)
html = response.text
result = re.findall('"id":"(\d{4,6})","num"', html, re.S)
for i in range(len(result)):
url_image = "http://ossweb-img.qq.com/images/lol/web201310/skin/big" + result[i] + ".jpg"
res = requests.get(url=url_image, headers=headers).content
with open("images" + str(result[i]) + ".jpg", "wb") as ob:
ob.write(res)
ob.close()
print(str(result[i]) + "is Done")
def main():
print('start the work')
url = 'http://lol.qq.com/biz/hero/champion.js'
get_hero(url)
if __name__ == '__main__':
main()
re.findall(‘if.?champion=(.?)data’, html, re.S)
re 正则模块很多地方都用到,很重要,长文详解python正则表达式这篇很详细,深入浅出,很棒。
compile函数、match函数、search函数、findall函数、finditer函数、split函数、sub函数、subn函数小结:
- 函数辨析:match和search的区别
re.match只匹配字符串的开始,如果字符串开始不符合正则表达式,则匹配失败,函数返回None;
re.search匹配整个字符串,直到找到一个匹配。- 函数辨析:3个匹配函数match、search、findall
match 和 search 只匹配一次 ,匹配不到返回None,findall 查找所有匹配结果。- 函数返回值
函数re.finditer 、 re.match和re.search 返回匹配对象,而findall、split返回列表。- re.compile函数是个谜。
对于小数据来说,re.search 比re.match 好用
re.split 在不同的数据提取里面相对有用
如之前要对2017-1-2 这样的时间序列做处理,只提取当中的月份‘1’,可以用split函数
dom = df[['Date', 'AQI']]
list1 = []
for j in dom['Date']:
time = j.split('-')[1] #对2017-1-2以 - 分割,然后取第1个数据
list1.append(time)
df['month'] = list1
time.sleep(t) 函数,t代表秒数
time.sleep(2) #推迟执行的秒数,应该是为了防止被反爬
拼图
代码非常多,有些细节也没仔细理解,大概的思路是先将要处理的目标图片分割成若干小图片,然后根据若干小图片的灰度值或RGB可以得到唯一的哈希值,然后去图片库中找哈希值近似的图片代替就可以了
但是这个代码有一个小问题,只能用jpg格式的图片,但很多时候我们处理的图片都是png格式的,这就需要我们手动转化一下,可以考虑用格式工厂转化。具体案例如下:
原图
处理后的马赛克图片
完整代码:
import os
import time
from functools import reduce
from threading import Thread
from PIL import Image
class MosaicMaker(object):
# 内部类,执行多线程拼图的任务类
class __SubTask:
def __init__(self, n, cur_sub_im, new_im, m, box):
self.n = n
self.cur_sub_im = cur_sub_im
self.new_im = new_im
self.m = m
self.box = box
def work(self):
# print("正在拼第%d张素材" % self.n)
# 计算key值(灰度值,平均RGB,hash值,三选一)
cur_sub_key = self.m.cal_key(self.cur_sub_im)
# 搜索最匹配图片(灰度值,平均RGB,hash值,三选一)
fit_sub = self.m.find_key(cur_sub_key)
self.new_im.paste(fit_sub, self.box)
# 内部类,执行多线程读取图库的任务类
class __ReadTask:
def __init__(self, n, full_path, fin_w, fin_h, m):
self.n = n
self.full_path = full_path
self.fin_w = fin_w
self.fin_h = fin_h
self.m = m
def read(self):
print("开始读取第%d张图片" % self.n)
cur = Image.open(self.full_path)
# 计算key值(灰度值,平均RGB,hash值,三选一)
key = self.m.cal_key(cur)
# 将素材缩放到目标大小
cur = cur.resize((self.fin_w, self.fin_h), Image.ANTIALIAS)
self.m.get_all_img().update({key: cur})
# 图库目录 目标文件 输出路径 子图尺寸 最小像素单位 拼图模式 默认尺寸
def __init__(self, db_path, aim_path, out_path, sub_width=64, sub_height=64, min_unit=5, mode="RGB", default_w=1600,
default_h=1280):
self.__db_path = db_path
self.__aim_path = aim_path
self.__out_path = out_path
self.__sub_width = sub_width
self.__sub_height = sub_height
self.__min_unit = min_unit
self.__mode = mode
self.__default_w = default_w
self.__default_h = default_h
self.__all_img = dict()
# 对外提供的接口
def make(self):
aim_im = Image.open(self.__aim_path)
aim_width = aim_im.size[0]
aim_height = aim_im.size[1]
print("计算子图尺寸")
if not self.__divide_sub_im(aim_width, aim_height):
print("使用默认尺寸")
aim_im = aim_im.resize((self.__default_w, self.__default_h), Image.ANTIALIAS)
aim_width = aim_im.size[0]
aim_height = aim_im.size[1]
print("读取图库")
start = time.time()
self.__read_all_img(self.__db_path, self.__sub_width, self.__sub_height)
print("耗时:%f秒" % (time.time() - start))
self.__core(aim_im, aim_width, aim_height)
def __core(self, aim_im, width, height):
new_im = Image.new("RGB", (width, height))
# 每行每列的图片数
w = width // self.__sub_width
print("源文件尺寸为:(w:%d h:%d)" % (width, height))
print("子图的尺寸为:(w:%d h:%d)" % (self.__sub_width, self.__sub_height))
print("w:%d" % w)
print("开始拼图,请稍等...")
start = time.time()
n = 1
thread_list = list()
for i in range(w):
task_list = list()
for j in range(w):
# 多线程版
left = i * self.__sub_width
up = j * self.__sub_height
right = (i + 1) * self.__sub_width
down = (j + 1) * self.__sub_height
box = (left, up, right, down)
cur_sub_im = aim_im.crop(box)
t = self.__SubTask(n, cur_sub_im, new_im, self, box)
task_list.append(t)
n += 1
thread = Thread(target=self.__sub_mission, args=(task_list,))
thread_list.append(thread)
for t in thread_list:
t.start()
for t in thread_list:
t.join()
print("拼图完成,共耗时%f秒" % (time.time() - start))
# 将原图与拼图合并,提升观感
new_im = Image.blend(new_im, aim_im, 0.35)
new_im.show()
new_im.save(self.__out_path)
# 拼图库线程执行的具体函数
@staticmethod
def __sub_mission(missions):
for task in missions:
task.work()
# 计算子图大小
def __divide_sub_im(self, width, height):
flag = True
g = self.__gcd(width, height)
if g < 20:
flag = False
width = self.__default_w
height = self.__default_h
g = 320
if g == width:
g = 320
self.__sub_width = self.__min_unit * (width // g)
self.__sub_height = self.__min_unit * (height // g)
return flag
# 读取全部图片,按(灰度值,平均RGB,hash值)保存 fin_w,fin_h素材最终尺寸
def __read_all_img(self, db_path, fin_w, fin_h):
files_name = os.listdir(db_path)
n = 1
# 开启5个线程加载图片
ts = list()
for i in range(5):
ts.append(list())
for file_name in files_name:
full_path = db_path + "\\" + file_name
if os.path.isfile(full_path):
read_task = self.__ReadTask(n, full_path, fin_w, fin_h, self)
ts[n % 5].append(read_task)
n += 1
tmp = list()
for i in ts:
t = Thread(target=self.__read_img, args=(i,))
t.start()
tmp.append(t)
for t in tmp:
t.join()
# 读取图库线程执行的具体函数
@staticmethod
def __read_img(tasks):
for task in tasks:
task.read()
# 计算key值
def cal_key(self, im):
if self.__mode == "RGB":
return self.__cal_avg_rgb(im)
elif self.__mode == "gray":
return self.__cal_gray(im)
elif self.__mode == "hash":
return self.__cal_hash(im)
else:
return ""
# 获取key值
def find_key(self, im):
if self.__mode == "RGB":
return self.__find_by_rgb(im)
elif self.__mode == "gray":
return self.__find_by_gray(im)
elif self.__mode == "hash":
return self.__find_by_hash(im)
else:
return ""
# 计算灰度值
@staticmethod
def __cal_gray(im):
if im.mode != "L":
im = im.convert("L")
return reduce(lambda x, y: x + y, im.getdata()) // (im.size[0] * im.size[1])
# 计算平均rgb值
@staticmethod
def __cal_avg_rgb(im):
if im.mode != "RGB":
im = im.convert("RGB")
pix = im.load()
avg_r, avg_g, avg_b = 0, 0, 0
n = 1
for i in range(im.size[0]):
for j in range(im.size[1]):
r, g, b = pix[i, j]
avg_r += r
avg_g += g
avg_b += b
n += 1
avg_r /= n
avg_g /= n
avg_b /= n
return str(avg_r) + "-" + str(avg_g) + "-" + str(avg_b)
# 计算hash
def __cal_hash(self, im):
im = im.resize((8, 8), Image.ANTIALIAS)
im = im.convert("L")
avg_gray = self.__cal_gray(im)
k = ""
_0 = "0"
_1 = "1"
for i in im.getdata():
if i < avg_gray:
k += _0
else:
k += _1
return k
# 辗转相除法求最大公约数
@staticmethod
def __gcd(a, b):
while a % b:
a, b = b, a % b
return b
# 获取最佳素材(按灰度)
def __find_by_gray(self, gray):
m = 255
k = 0
for key in self.__all_img.keys():
cur_dif = abs(key - gray)
if cur_dif < m:
k = key
m = cur_dif
return self.__all_img[k]
# 获取最佳素材(按pHash)
def __find_by_hash(self, sub_hash):
m = 65
k = 0
for key in self.__all_img.keys():
cur_dif = self.__dif_num(sub_hash, key)
if cur_dif < m:
k = key
m = cur_dif
return self.__all_img[k]
@staticmethod
def __dif_num(hash1, hash2):
n = 0
for i in range(64):
if hash1[i] != hash2[i]:
n += 1
return n
# # 获取最佳素材(按平均rgb)
def __find_by_rgb(self, sub_rgb):
sub_r, sub_g, sub_b = sub_rgb.split("-")
m = 255
k = ""
for key in self.__all_img.keys():
src_r, src_g, src_b = key.split("-")
cur_dif = abs(float(sub_r) - float(src_r)) + abs(float(sub_g) - float(src_g)) + abs(
float(sub_b) - float(src_b))
if cur_dif < m:
m = cur_dif
k = key
return self.__all_img[k]
def get_all_img(self):
return self.__all_img
if __name__ == '__main__':
m = MosaicMaker("D:\\思维导图\\LOL英雄图片", r"C:\Users\hao\Desktop\编程\趣味案例\**.jpg",
"方形3.jpg")
m.make()
pass