如何用cv2快速检测IP摄像头的连接状态？

问题描述 #

我有一套可在2台IP摄像头上运行的代码，在执行主算法前需要检测IP摄像头是否已连接。此前用于检测USB摄像头的代码运行正常，但切换到IP摄像头后遇到了超时问题：

import cv2 

def check_connections():
    # 检测摄像头1
    cam1_ok = False    
    cam1 = cv2.VideoCapture('rtsp://admin:Tom12345.@192.168.0.33:554/onvif1') 
    if cam1.isOpened():
        cam1_ok = True

    # 检测摄像头2
    cam2_ok = False
    cam2 = cv2.VideoCapture('rtsp://admin:Tom12345.@192.168.0.33:554/onvif1') 
    if cam2.isOpened():
        cam2_ok = True

    print(f'Cam1 Connection Status: {cam1_ok}')
    print(f'Cam2 Connection Status: {cam2_ok}')


check_connections()

当摄像头正常连接时，能正确返回Cam1 Connection Status: True和Cam2 Connection Status: True；但摄像头断开时，cv2.VideoCapture()需要耗时3分钟才会返回False，而USB摄像头检测能即时出结果。尝试过try...except语句未能解决，求更快的检测方法。

优化方案 #

方法1：设置OpenCV的RTSP超时参数 #

OpenCV的VideoCapture支持通过CAP_PROP系列参数缩短RTSP流的超时时间，同时结合读取帧的操作验证连接（仅用isOpened()可能出现误判）：

import cv2

def check_cam_connection(rtsp_url, timeout=5):
    cam = cv2.VideoCapture(rtsp_url)
    # 设置打开流和读取帧的超时时间（单位：毫秒）
    cam.set(cv2.CAP_PROP_OPEN_TIMEOUT_MSEC, timeout * 1000)
    cam.set(cv2.CAP_PROP_READ_TIMEOUT_MSEC, timeout * 1000)
    
    # 尝试读取一帧验证连接有效性
    ret, _ = cam.read()
    cam_ok = ret and cam.isOpened()
    
    cam.release()
    return cam_ok

def check_connections():
    # 注意：原代码两台摄像头用了同一个URL，疑似笔误，此处保留示例格式
    cam1_url = 'rtsp://admin:Tom12345.@192.168.0.33:554/onvif1'
    cam2_url = 'rtsp://admin:Tom12345.@192.168.0.33:554/onvif1'
    
    cam1_ok = check_cam_connection(cam1_url, timeout=5)
    cam2_ok = check_cam_connection(cam2_url, timeout=5)
    
    print(f'Cam1 Connection Status: {cam1_ok}')
    print(f'Cam2 Connection Status: {cam2_ok}')

check_connections()

方法2：多线程并行检测 #

如果需要同时检测多个摄像头，用多线程避免串行等待超时，提升整体检测效率：

import cv2
import threading

def check_cam_thread(rtsp_url, result_dict, cam_name, timeout=5):
    cam = cv2.VideoCapture(rtsp_url)
    cam.set(cv2.CAP_PROP_OPEN_TIMEOUT_MSEC, timeout * 1000)
    cam.set(cv2.CAP_PROP_READ_TIMEOUT_MSEC, timeout * 1000)
    
    ret, _ = cam.read()
    result_dict[cam_name] = ret and cam.isOpened()
    
    cam.release()

def check_connections():
    cam_info = {
        'Cam1': 'rtsp://admin:Tom12345.@192.168.0.33:554/onvif1',
        'Cam2': 'rtsp://admin:Tom12345.@192.168.0.34:554/onvif1'  # 修正为不同摄像头URL
    }
    result = {}
    threads = []
    
    for name, url in cam_info.items():
        thread = threading.Thread(target=check_cam_thread, args=(url, result, name, 5))
        threads.append(thread)
        thread.start()
    
    for thread in threads:
        thread.join()
    
    for name, status in result.items():
        print(f'{name} Connection Status: {status}')

check_connections()

方法3：先通过网络Ping检测连通性 #

如果摄像头支持ICMP协议，先Ping检测设备是否在线，再尝试打开RTSP流，进一步缩短无效检测的耗时：

import os
import cv2

def ping_host(host):
    # 跨平台Ping命令适配
    param = '-n' if os.name == 'nt' else '-c'
    command = ['ping', param, '1', '-w', '2000', host]
    return os.system(command) == 0

def check_cam_connection(rtsp_url, timeout=5):
    # 从RTSP地址中提取摄像头IP
    host = rtsp_url.split('@')[1].split(':')[0]
    # 设备不在线直接返回False
    if not ping_host(host):
        return False
    
    cam = cv2.VideoCapture(rtsp_url)
    cam.set(cv2.CAP_PROP_OPEN_TIMEOUT_MSEC, timeout * 1000)
    cam.set(cv2.CAP_PROP_READ_TIMEOUT_MSEC, timeout * 1000)
    
    ret, _ = cam.read()
    cam_ok = ret and cam.isOpened()
    
    cam.release()
    return cam_ok

# 后续调用逻辑同方法1

注意：部分摄像头可能禁用了ICMP协议，这种情况下该方法不适用。

内容的提问来源于stack exchange，提问作者Barış Aktaş