|
使用到的第三方库
Package Version
--------------- ---------
baidu-aip 2.2.18.0
jieba 0.42.1
moviepy 1.0.3
numpy 1.20.2
opencv-python 4.5.1.48
Pillow 8.2.0
requests 2.25.1
wordcloud 1.8.1
you-get 0.4.1520
B站弹幕爬取
思路
通过视频BV号请求cid，再使用cid请求弹幕文件，最后使用正则表达式去匹配弹幕文本，将匹配出来的结果保存在本地供之后使用，代码及思路比较简单，就不做过多赘述
实现
cid请求链接：https://api.bilibili.com/x/web-interface/view?bvid=
弹幕请求链接：https://api.bilibili.com/x/v1/dm/list.so?oid=
参考代码
def get_cid(cls, bv):
url = "https://api.bilibili.com/x/web-interface/view?bvid=" + str(bv)
response = requests.get(url)
dirt = json.loads(response.text)
aid = dirt['data']['cid']
return str(aid)
def get_barrage(cls, bv, to_file_path):
headers = {
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/71.0.3578.98 Safari/537.36",
}
cid = cls.get_cid(bv)
response = requests.get("https://api.bilibili.com/x/v1/dm/list.so?oid=" + cid, headers=headers)
html_doc = response.content.decode('utf-8')
regex = re.compile("<d.*?>(.*?)</d>")
DanMu = regex.findall(html_doc)
with open(to_file_path, "w", encoding="utf_8")as f:
for i in DanMu:
f.write(i)
f.write("\n")
视频下载
思路
使用第三方开源库you-get进行下载
you-get支持的音视频网站
| Site |
URL |
Videos? |
Images? |
Audios? |
| YouTube |
https://www.youtube.com/ |
✓ |
|
|
| Twitter |
https://twitter.com/ |
✓ |
✓ |
|
| VK |
http://vk.com/ |
✓ |
✓ |
|
| Vine |
https://vine.co/ |
✓ |
|
|
| Vimeo |
https://vimeo.com/ |
✓ |
|
|
| Veoh |
http://www.veoh.com/ |
✓ |
|
|
| Tumblr |
https://www.tumblr.com/ |
✓ |
✓ |
✓ |
| TED |
http://www.ted.com/ |
✓ |
|
|
| SoundCloud |
https://soundcloud.com/ |
|
|
✓ |
| SHOWROOM |
https://www.showroom-live.com/ |
✓ |
|
|
| Pinterest |
https://www.pinterest.com/ |
|
✓ |
|
| MTV81 |
http://www.mtv81.com/ |
✓ |
|
|
| Mixcloud |
https://www.mixcloud.com/ |
|
|
✓ |
| Metacafe |
http://www.metacafe.com/ |
✓ |
|
|
| Magisto |
http://www.magisto.com/ |
✓ |
|
|
| Khan Academy |
https://www.khanacademy.org/ |
✓ |
|
|
| Internet Archive |
https://archive.org/ |
✓ |
|
|
| Instagram |
https://instagram.com/ |
✓ |
✓ |
|
| InfoQ |
http://www.infoq.com/presentations/ |
✓ |
|
|
| Imgur |
http://imgur.com/ |
|
✓ |
|
| Heavy Music Archive |
http://www.heavy-music.ru/ |
|
|
✓ |
| Freesound |
http://www.freesound.org/ |
|
|
✓ |
| Flickr |
https://www.flickr.com/ |
✓ |
✓ |
|
| FC2 Video |
http://video.fc2.com/ |
✓ |
|
|
| Facebook |
https://www.facebook.com/ |
✓ |
|
|
| eHow |
http://www.ehow.com/ |
✓ |
|
|
| Dailymotion |
http://www.dailymotion.com/ |
✓ |
|
|
| Coub |
http://coub.com/ |
✓ |
|
|
| CBS |
http://www.cbs.com/ |
✓ |
|
|
| Bandcamp |
http://bandcamp.com/ |
|
|
✓ |
| AliveThai |
http://alive.in.th/ |
✓ |
|
|
| interest.me |
http://ch.interest.me/tvn |
✓ |
|
|
| 755 ナナゴーゴー |
http://7gogo.jp/ |
✓ |
✓ |
|
| niconico ニコニコ動画 |
http://www.nicovideo.jp/ |
✓ |
|
|
| 163 网易视频 网易云音乐 |
http://v.163.com/ http://music.163.com/ |
✓ |
|
✓ |
| 56网 |
http://www.56.com/ |
✓ |
|
|
| AcFun |
http://www.acfun.cn/ |
✓ |
|
|
| Baidu 百度贴吧 |
http://tieba.baidu.com/ |
✓ |
✓ |
|
| 爆米花网 |
http://www.baomihua.com/ |
✓ |
|
|
| bilibili 哔哩哔哩 |
http://www.bilibili.com/ |
✓ |
✓ |
✓ |
| 豆瓣 |
http://www.douban.com/ |
✓ |
|
✓ |
| 斗鱼 |
http://www.douyutv.com/ |
✓ |
|
|
| 凤凰视频 |
http://v.ifeng.com/ |
✓ |
|
|
| 风行网 |
http://www.fun.tv/ |
✓ |
|
|
| iQIYI 爱奇艺 |
http://www.iqiyi.com/ |
✓ |
|
|
| 激动网 |
http://www.joy.cn/ |
✓ |
|
|
| 酷6网 |
http://www.ku6.com/ |
✓ |
|
|
| 酷狗音乐 |
http://www.kugou.com/ |
|
|
✓ |
| 酷我音乐 |
http://www.kuwo.cn/ |
|
|
✓ |
| 乐视网 |
http://www.le.com/ |
✓ |
|
|
| 荔枝FM |
http://www.lizhi.fm/ |
|
|
✓ |
| 懒人听书 |
http://www.lrts.me/ |
|
|
✓ |
| 秒拍 |
http://www.miaopai.com/ |
✓ |
|
|
| MioMio弹幕网 |
http://www.miomio.tv/ |
✓ |
|
|
| MissEvan 猫耳FM |
http://www.missevan.com/ |
|
|
✓ |
| 痞客邦 |
https://www.pixnet.net/ |
✓ |
|
|
| PPTV聚力 |
http://www.pptv.com/ |
✓ |
|
|
| 齐鲁网 |
http://v.iqilu.com/ |
✓ |
|
|
| QQ 腾讯视频 |
http://v.qq.com/ |
✓ |
|
|
| 企鹅直播 |
http://live.qq.com/ |
✓ |
|
|
| Sina 新浪视频 微博秒拍视频 |
http://video.sina.com.cn/ http://video.weibo.com/ |
✓ |
|
|
| Sohu 搜狐视频 |
http://tv.sohu.com/ |
✓ |
|
|
| Tudou 土豆 |
http://www.tudou.com/ |
✓ |
|
|
| 阳光卫视 |
http://www.isuntv.com/ |
✓ |
|
|
| Youku 优酷 |
http://www.youku.com/ |
✓ |
|
|
| 战旗TV |
http://www.zhanqi.tv/lives |
✓ |
|
|
| 央视网 |
http://www.cntv.cn/ |
✓ |
|
|
| Naver 네이버 |
http://tvcast.naver.com/ |
✓ |
|
|
| 芒果TV |
http://www.mgtv.com/ |
✓ |
|
|
| 火猫TV |
http://www.huomao.com/ |
✓ |
|
|
| 阳光宽频网 |
http://www.365yg.com/ |
✓ |
|
|
| 西瓜视频 |
https://www.ixigua.com/ |
✓ |
|
|
| 新片场 |
https://www.xinpianchang.com/ |
✓ |
|
|
| 快手 |
https://www.kuaishou.com/ |
✓ |
✓ |
|
| 抖音 |
https://www.douyin.com/ |
✓ |
|
|
| TikTok |
https://www.tiktok.com/ |
✓ |
|
|
| 中国体育(TV) |
http://v.zhibo.tv/ http://video.zhibo.tv/ |
✓ |
|
|
| 知乎 |
https://www.zhihu.com/ |
✓ |
|
|
# 获取视频信息
you-get -i https://www.bilibili.com/video/BV1f4411M7QC
# 下载视频
you-get --format=flv -o E:\Desktop\output https://www.bilibili.com/video/BV1f4411M7QC
视频、音频剪辑和音频提取
思路
这部分的需求非常简单，就是剪下视频或者音频中的某一段并保存
Python有一个叫moviepy的第三方库，可以实现视频的剪辑、拼接，音频的剪辑、拼接、提取，以及音视频的合并等操作
参考代码
def cut_video(cls, origin_file_path, to_file_path, start, end):
"""
视频剪辑
:param origin_file_path: 原视频文件路径
:param to_file_path: 保存路径
:param start: 起始时间点
:param end: 结束时间点
"""
clip = VideoFileClip(origin_file_path).subclip(start, end)
clip.write_videofile(to_file_path)
def cut_audio(cls, origin_file_path, to_file_path, start, end):
"""
音频剪辑
:param origin_file_path: 原视频文件路径
:param to_file_path: 保存路径
:param start: 起始时间点
:param end: 结束时间点
"""
clip = AudioFileClip(origin_file_path).subclip(start, end)
clip.write_audiofile(to_file_path)
def get_audio_from_video(cls, video_file_path, to_file_path):
"""
音频提取
:param video_file_path: 视频文件路径
:param to_file_path: 音频文件路径
"""
video = VideoFileClip(video_file_path)
video.audio.write_audiofile(to_file_path)
视频帧提取
思路
使用opencv-python(cv2)打开视频文件并按帧读取，再将每一帧保存到文件夹中
参考代码
def split(cls, from_file_path, to_folder_path, frames=0):
"""
视频按帧读取并保存
:param from_file_path: 视频路径
:param to_folder_path: 保存路径
:param frames: 保存帧数(张数)，为0则保存所有帧
"""
vc = cv2.VideoCapture(from_file_path) # cv2打开视频文件
frames_count = vc.get(7) # 获取视频总帧数
c = 0
if vc.isOpened():
ret, frame = vc.read() # 按帧读取视频
else:
ret = False
while ret:
if 0 < frames == c:
break
ret, frame = vc.read() # 读取每一视频帧，并保存至图片中
cv2.imwrite(os.path.join(to_folder_path, '{}.jpg'.format(c)), frame)
c += 1
if c == frames_count - 1:
break
print('第 {} 张图片存放成功！'.format(c))
图片二值化
思路
图片二值化这里有两种思路，一种是使用opencv，还有一种方法是使用百度智能云的人像分割接口。
两种方法各有优劣：
- 使用opencv的速度快，但是只能对整张图片二值化，无法有效提取出图片主体部分，只适用于纯色背景及轮廓分明的图片，当图片中有背景或者其他干扰画面时，效果不理想，达不到做词云遮罩的效果
- 百度的人像分割接口可以将图片中的人物抠出来，单独对人物进行二值化，但是速度很慢(处理速度慢，还限制接口并发数)，一千张图片往往需要一两个小时
所以具体使用时需要根据视频的情况进行切换
下面为两周处理方法的不同效果(图一为cv2，图二为百度人像分割)
参考代码
def binary_option_cv2(cls, from_file_path, to_file_path):
"""
图片二值化并保存(使用cv2)
:param from_file_path: 原图路径
:param to_file_path: 二值化图路径
"""
img = cv2.imread(from_file_path)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
new_gray = np.uint8((255 * (gray / 255.0) ** 1.4))
dst = cv2.adaptiveThreshold(new_gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 15, 1)
cv2.medianBlur(dst, 5)
cv2.imwrite(to_file_path, dst)
def binary_option_baidu(cls, from_file_path, to_file_path):
"""
图片二值化并保存(使用百度人像分割)
:param from_file_path: 原图路径
:param to_file_path: 二值化图路径
"""
def get_file_content(filePath):
with open(filePath, 'rb') as fp:
return fp.read()
height, width, bgr = cv2.imread(from_file_path).shape
image = get_file_content(from_file_path)
cls.client.bodySeg(image)
res = cls.client.bodySeg(image)
labelmap = base64.b64decode(res['labelmap'])
labelimg = np.frombuffer(labelmap, np.uint8) # 转化为np数组 0-255
labelimg = cv2.imdecode(labelimg, 1)
labelimg = cv2.resize(labelimg, (width, height), interpolation=cv2.INTER_NEAREST)
img_new = np.where(labelimg == 1, 255, labelimg) # 将 1 转化为 255
cv2.imwrite(to_file_path, img_new)
词云图片生成
思路
使用wordcloud库，并使用前面爬取的B站弹幕作为词云内容，二值化图片作为遮罩
原图与词云图拼接和图片合并生成视频
思路
使用numpy拼接图片，使用cv2将拼接的图片写入视频流并保存
为了将视频与音轨对齐，生成视频时需要设置合适的视频帧率（与原视频保持一致），原视频帧率可以使用播放器查看，也可以使用cv2获取
参考代码
def joint(cls, origin_folder, word_cloud_folder, to_file_path):
"""
批量拼接图片并合成视频
:param origin_folder: 原图文件夹
:param word_cloud_folder: 词云图片文件夹
:param to_file_path: 保存路径
"""
num_list = [int(str(i).split('.')[0]) for i in os.listdir(origin_folder)]
fps = 30 # 视频帧率，需要根据原视频帧率做调整
height, width, _ = cv2.imread(os.path.join(origin_folder, '{}.jpg'.format(num_list[0]))).shape # 视频高度和宽度
width = width * 2
# 创建一个写入操作;
video_writer = cv2.VideoWriter(to_file_path, cv2.VideoWriter_fourcc(*'mp4v'), fps, (width, height))
for i in sorted(num_list):
i = '{}.jpg'.format(i)
ori_jpg = os.path.join(origin_folder, str(i))
word_jpg = os.path.join(word_cloud_folder, str(i))
# com_jpg = os.path.join(Composite_path,str(i))
ori_arr = cv2.imread(ori_jpg)
word_arr = cv2.imread(word_jpg)
# 利用 Numpy 进行拼接
com_arr = np.hstack((ori_arr, word_arr))
video_writer.write(com_arr) # 将每一帧画面写入视频流中
print("{}写入视频流成功".format(ori_jpg))
音视频合并和视频导出
思路
与前面 原图与词云图拼接和图片合并生成视频 思路相似
参考代码
def set_audio_for_video(cls, video_file_path, audio_file_path, to_file_path):
"""
音视频合并
:param video_file_path: 视频文件路径
:param audio_file_path: 音频文件路径
:param to_file_path: 保存路径
"""
video = VideoFileClip(video_file_path)
audio = AudioFileClip(audio_file_path)
new_video = video.set_audio(audio)
new_video.write_videofile(to_file_path)
最终效果
到此这篇关于用python制作词云视频详解的文章就介绍到这了,希望对大家有帮助，更多相关python视频请搜索社区以前的文章或继续浏览下面的相关文章,希望大家以后多多支持社区！ | 
转播
