如何使用OpenCV独立控制立体相机的两个摄像头？

我有一台可正常作为USB摄像头工作的立体相机，希望独立控制它的两个摄像头模组，用于相机标定等立体视觉相关操作。我尝试用OpenCV通过相机索引实现该功能：笔记本内置摄像头索引为0，USB立体相机索引为1。当传入索引[0,1]时，能成功调用笔记本摄像头与立体相机；但传入索引[1,2]时，出现如下报错：

[ERROR:0@0.125] global obsensor_uvc_stream_channel.cpp:159 cv::obsensor::getStreamChannelGroup Camera index out of range
[ERROR:0@0.163] global obsensor_uvc_stream_channel.cpp:159 cv::obsensor::getStreamChannelGroup Camera index out of range
Error: One or both camera streams could not be opened.

以下是我使用的代码：

import cv2
def capture_dual_camera_module(camera_index_left=1, camera_index_right=2):
    """
    Captures video streams from two cameras (stereo camera) simultaneously.
    Args:
        camera_index_left (int, optional): Index of the left camera (default: 0).
        camera_index_right (int, optional): Index of the right camera (default: 1).
    Returns:
        None
    """
    cap_left = cv2.VideoCapture(camera_index_left)
    cap_right = cv2.VideoCapture(camera_index_right)
    if not (cap_left.isOpened() and cap_right.isOpened()):
        print("Error: One or both camera streams could not be opened.")
        return
    while True:
        # Capture frames from both cameras
        ret_left, frame_left = cap_left.read()
        ret_right, frame_right = cap_right.read()
        if not ret_left or not ret_right:
            print("Error: Couldn't read frames from one or both camera streams.")
            break
        # Display the resulting frames (adjust window names for clarity):
        cv2.imshow('Left Camera', frame_left)
        cv2.imshow('Right Camera', frame_right)
        # Break the loop on 'q' key press
        if cv2.waitKey(1) & 0xFF == ord('q'):
            break
    # Release capture resources
    cap_left.release()
    cap_right.release()
    cv2.destroyAllWindows()
if __name__ == "__main__":
    capture_dual_camera_module()  

我使用的相机外观如下：

这类一体化立体相机在系统里通常被识别为单个设备，不是两个独立的USB摄像头，所以没法用两个不同的索引（1和2）分别调用左右模组。下面是几种可行的解决方法：

1. 尝试UVC多流通道切换 #

部分立体相机支持UVC多流输出，单个设备可同时输出左右摄像头的视频流。这时不需要换设备索引，而是通过设置CAP_PROP_CHANNEL属性切换通道，代码修改如下：

import cv2

def capture_stereo_camera(camera_index=1):
    # 打开同一个立体相机设备
    cap_left = cv2.VideoCapture(camera_index)
    cap_right = cv2.VideoCapture(camera_index)
    
    # 设置左右通道（不同相机的通道编号可能是0/1或1/2，自行测试）
    cap_left.set(cv2.CAP_PROP_CHANNEL, 0)
    cap_right.set(cv2.CAP_PROP_CHANNEL, 1)
    
    if not (cap_left.isOpened() and cap_right.isOpened()):
        print("Error: Failed to open stereo camera streams.")
        return
    
    while True:
        ret_left, frame_left = cap_left.read()
        ret_right, frame_right = cap_right.read()
        
        if not ret_left or not ret_right:
            print("Error: Failed to read frames.")
            break
        
        cv2.imshow('Left Module', frame_left)
        cv2.imshow('Right Module', frame_right)
        
        if cv2.waitKey(1) & 0xFF == ord('q'):
            break
    
    cap_left.release()
    cap_right.release()
    cv2.destroyAllWindows()

if __name__ == "__main__":
    capture_stereo_camera()

2. 使用厂商专属SDK #

如果相机不支持UVC多流，直接用厂商提供的SDK是最靠谱的选择——比如常见的Intel RealSense、ZED等立体相机都有专属SDK，能直接获取左右摄像头的独立帧，还自带标定、深度计算等功能，比OpenCV适配性强得多。

3. 检查系统设备列表 #

先确认系统识别到的摄像头数量：

• Windows：打开「设备管理器→图像设备」查看
• Linux：终端执行v4l2-ctl --list-devices
• macOS：终端执行system_profiler SPCameraDataType
  如果立体相机只显示为一个设备，那肯定没法通过两个索引调用。

4. 单流获取后分割帧 #

如果相机输出的是左右拼接的画面（宽度是单摄像头的两倍），可以获取单帧后手动分割成左右两部分：

import cv2

def split_stereo_frame(frame):
    height, width = frame.shape[:2]
    left_frame = frame[:, :width//2]
    right_frame = frame[:, width//2:]
    return left_frame, right_frame

def capture_and_split(camera_index=1):
    cap = cv2.VideoCapture(camera_index)
    
    if not cap.isOpened():
        print("Error: Failed to open camera.")
        return
    
    while True:
        ret, frame = cap.read()
        if not ret:
            print("Error: Failed to read frame.")
            break
        
        left_frame, right_frame = split_stereo_frame(frame)
        
        cv2.imshow('Left Module', left_frame)
        cv2.imshow('Right Module', right_frame)
        
        if cv2.waitKey(1) & 0xFF == ord('q'):
            break
    
    cap.release()
    cv2.destroyAllWindows()

if __name__ == "__main__":
    capture_and_split()

内容的提问来源于stack exchange，提问作者Ameer Hamzah