使用node-red+opencv+mqtt实现相机图像云端查看

一、本机环境

jetson nano ubuntu20.04

python 3.12

pip install opencv-python paho-mqtt

安装并启动mqtt broker

# 安装 Mosquitto
sudo apt update
sudo apt install mosquitto mosquitto-clients -y

# 启动并设置开机自启
sudo systemctl enable mosquitto
sudo systemctl start mosquitto

# 验证是否运行
sudo systemctl status mosquitto

二、制作方法

mqtt broker启动成功后，执行这个python脚本

# script.py
import cv2
import paho.mqtt.client as mqtt
import base64
import time

# MQTT 配置
MQTT_BROKER = "localhost"
MQTT_PORT = 1883
MQTT_TOPIC = "camera/image"

# 打开摄像头
cap = cv2.VideoCapture(0)
if not cap.isOpened():
    print("无法打开摄像头")
    exit()

# 指定 callback_api_version
client = mqtt.Client(callback_api_version=mqtt.CallbackAPIVersion.VERSION2)

# 连接 MQTT
try:
    client.connect(MQTT_BROKER, MQTT_PORT, 60)
    client.loop_start()  # 启动后台线程处理网络通信
except Exception as e:
    print(f"MQTT 连接失败: {e}")
    exit()

print("摄像头启动，开始发布图像...")

try:
    while True:
        ret, frame = cap.read()
        if not ret:
            print("摄像头读取失败")
            break

        # 缩放图像（减小带宽）
        frame = cv2.resize(frame, (640, 480))

        # 编码为 JPEG
        _, buffer = cv2.imencode('.jpg', frame, [cv2.IMWRITE_JPEG_QUALITY, 80])
        
        # 转为 Base64 字符串
        jpg_as_text = base64.b64encode(buffer).decode('utf-8')

        # 发布到 MQTT
        result = client.publish(MQTT_TOPIC, jpg_as_text, qos=0)
        
        # 可选：检查是否发布成功
        # if result.rc != mqtt.MQTT_ERR_SUCCESS:
        #     print("发布失败")

        time.sleep(0.1)  # 控制帧率（约10fps）

except KeyboardInterrupt:
    print("\n停止发布")

finally:
    cap.release()
    client.loop_stop()  # 停止后台线程
    client.disconnect()

在node-red端则需要两个节点，一个是mqtt_in 和 ui_template节点，mqtt_in节点需要配置与python端一致的话题和端口

ui_template 则需要输入以下代码：

<div style="text-align: center;">
  <img ng-src="data:image/jpeg;base64,{{msg.payload}}" 
       style="max-width: 100%; height: auto; border: 1px solid #ccc; border-radius: 8px;"
       alt="Camera Stream">
  <p style="font-size: 12px; color: #666;">Last update: {{ $flow.lastUpdate }}</p>
</div>

<script>
// 更新时间戳
(function(scope) {
    scope.$watch('msg', function(msg) {
        if (msg) {
            scope.$flow.lastUpdate = new Date().toLocaleTimeString();
        }
    });
})(scope);
</script>

部署后，mqtt_in连接上服务器就对了

然后输入 127.0.0.1:1880/ui 进入ui界面，即可查看相机图像，也可通过远程ip访问查看

可以调整python端控制帧率的time.sleep函数，达到实时传输