Ubuntu 22.04下无法获取DJI Tello无人机实时视频流求助

问题情况 #

代码在Windows系统可正常实现Tello无人机实时视频流显示，但在Ubuntu 22.04.3 LTS运行时，OpenCV窗口无法弹出，控制台输出H264解码相关错误。

控制台错误输出 #

base) nirbhay@nirbhay:~/Desktop/Tello_code$ python LiveStream.py 
[INFO] tello.py - 129 - Tello instance was initialized. Host: '192.168.10.1'. Port: '8889'.
[INFO] tello.py - 438 - Send command: 'command'
[INFO] tello.py - 462 - Response command: 'ok'
Drone Battery Percentage:
100
Live Stream Address:
udp://@0.0.0.0:11111
[INFO] tello.py - 438 - Send command: 'streamon'
[INFO] tello.py - 462 - Response streamon: 'ok'
non-existing PPS 0 referenced
non-existing PPS 0 referenced
decode_slice_header error
no frame!
non-existing PPS 0 referenced
non-existing PPS 0 referenced
decode_slice_header error
no frame!
non-existing PPS 0 referenced
non-existing PPS 0 referenced
decode_slice_header error
no frame!
non-existing PPS 0 referenced
non-existing PPS 0 referenced
decode_slice_header error
no frame!
non-existing PPS 0 referenced
non-existing PPS 0 referenced
decode_slice_header error
no frame!
non-existing PPS 0 referenced
non-existing PPS 0 referenced
decode_slice_header error
no frame!
error while decoding MB 22 30, bytestream 127
error while decoding MB 58 37, bytestream -6

原代码 #

from djitellopy import tello
import cv2
import time
#import libh264decoder

drone = tello.Tello()
drone.connect()
print("Drone Battery Percentage:")
print(drone.get_battery())
print("Live Stream Address:")
print(drone.get_udp_video_address())

drone.streamon()

while True:
    img=drone.get_frame_read().frame
    #print("image:",img)
    img=cv2.resize(img,(660,420))
    cv2.imshow("Image",img)
    #time.sleep(3)
    cv2.waitKey(1)

    #time.sleep(1)

依赖与环境 #

• Python包版本：

  djitellopy                    2.5.0
  opencv-python                 4.9.0.80
  

• 操作系统：Ubuntu 22.04.3 LTS

解决步骤 #

1. 安装系统级H264解码库 #

Ubuntu下OpenCV依赖系统解码器，缺失会导致H264流解码失败，执行以下命令安装：

sudo apt update && sudo apt install libavcodec-dev libavformat-dev libswscale-dev libh264decoder-dev

2. 修改代码修复解码与帧处理逻辑 #

启用libh264decoder并优化帧读取逻辑，避免空帧导致窗口无法渲染：

from djitellopy import tello
import cv2
import time
import libh264decoder  # 取消注释启用内置解码器

drone = tello.Tello()
drone.connect()
print("Drone Battery Percentage:")
print(drone.get_battery())
print("Live Stream Address:")
print(drone.get_udp_video_address())

drone.streamon()
frame_read = drone.get_frame_read()  # 提前获取帧读取对象，避免循环内重复初始化

while True:
    # 检查流是否停止
    if frame_read.stopped:
        frame_read.stop()
        break
    img = frame_read.frame
    # 跳过空帧，避免OpenCV报错
    if img is None:
        continue
    img = cv2.resize(img, (660, 420))
    cv2.imshow("Image", img)
    # 添加窗口关闭逻辑，按q退出
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

# 清理资源
drone.streamoff()
cv2.destroyAllWindows()

3. 强制指定OpenCV窗口后端（可选） #

如果窗口仍不显示，强制使用GTK或QT后端，在代码开头添加：

# 强制使用GTK后端
cv2.namedWindow("Image", cv2.WINDOW_GTK_NORMAL)
# 或者使用QT后端
# cv2.namedWindow("Image", cv2.WINDOW_QT_NORMAL)

4. 开放UDP端口（若启用防火墙） #

如果Ubuntu开启了ufw防火墙，允许UDP 11111端口的入站流量：

sudo ufw allow 11111/udp

原因说明 #

控制台的解码错误是H264流解码失败的典型表现，Linux下OpenCV依赖系统级解码器，而djitellopy默认逻辑在Linux下对空帧处理不完善，导致OpenCV无法正常渲染窗口。通过补充系统解码器、优化帧读取逻辑，即可解决问题。

内容的提问来源于stack exchange，提问作者NIrbhay Mathur