python - cv2.imdecode()从base64中的图像(通过Websocket收到的mimetype image/jpeg)返回None

Question

我使用websockets接收视频帧。图像以base64 mimetype image / jpeg编码。我正在尝试将图像转换为np.ndarray。当我阅读图像文件代码时，它工作正常。但是，当我尝试从套接字流中读取图像时，出现了问题。

        image_data = base64.b64decode(part.encoded_image)
        np_array = np.frombuffer(image_data, np.uint8)
        image = cv2.imdecode(np_array, cv2.IMREAD_UNCHANGED)

在docs中，如果图片不足或损坏，cv2.imdecode()返回None。我的图像是高清格式1280/720。数据没有损坏，因为我可以使用在线转换器成功解码。
这是接收到的数据的开始:

data:image/jpeg;base64,/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAoHBwgHBgoICAgLCgoLDhgQDg0NDh0VFhEYIx8lJCIfIiEmKzcvJik0KSEiMEExNDk7Pj4+JS5ESUM8SDc9Pjv/

Answer 1

您必须跳过data:image/jpeg;base64,以获得仅base64数据，它将为您提供正确的图像数据，并且cv2将对其进行解码和显示。
在示例中，我为此使用split(',', 1)-它为我提供了data:image/jpeg;base64,变量中的header和为data中的图像数据。
我在代码中使用完整图像。

import base64
import numpy as np
import cv2

encoded_image = 'data:image/jpeg;base64,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'

header, data = encoded_image.split(',', 1)
#print('header:', header)
#print('  data:', data[:20])

image_data = base64.b64decode(data)
#print('result:', image_data[:20])

np_array = np.frombuffer(image_data, np.uint8)
#print(' array:', np_array[:2])

image = cv2.imdecode(np_array, cv2.IMREAD_UNCHANGED)
#print(' image:', image[:2])

cv2.imshow('image', image)
cv2.waitKey(0)

cv2.destroyAllWindows()

结果:

(尺寸为7x7的人脸图像)