opencv多线程显示的问题和解决方法

超、凢脫俗 2023-05-22 15:28 6阅读 0赞

## opencv多线程显示的问题和解决方法 ##

--------------------

## 1.问题描述 ##

由于业务需求,需要在多线程中,使用OpenCV读取图片并显示,实现很简单,只需要调用python的threading创建线程,并在线程中显示图片即可,但会出现一个问题: 只能显示第一张图图片,第二张图图片就会阻塞掉!!!

这是最初的精简的BUG代码

import os
    import cv2
    import glob
    import threading
    
    
    class DisplayTask(object):
        """
        """
    
        def __init__(self, ):
            """
            :param width:
            :param height:
            """
            self.win_name = "frame"
    
        def show(self, image_path):
            print("block here 1")
            frame = cv2.imread(image_path)
            print("block here 2")
            cv2.imshow(self.win_name, frame)
            print("block here 3")
            cv2.waitKey(1000)
            cv2.destroyAllWindows()
    
    
    def display_task(th_id, image_path):
        print("start thread-{}".format(th_id))
        dt = DisplayTask()
        dt.show(image_path)
    
    
    def demo(image_dir):
        image_list = glob.glob(os.path.join(image_dir, "*.jpg"))
        print("have image:{}".format(len(image_list)))
        for th_id, image_path in enumerate(image_list):
            thread = threading.Thread(target=display_task, args=(th_id, image_path,))
            thread.start()
            # 等待线程执行完毕
            thread.join()
            print("finish thread-{}".format(th_id))
    
    
    if __name__ == "__main__":
        image_dir = "/media/dm/dm1/git/python-learning-notes/data"
        demo(image_dir)

运行结果:

> have image:4  
> start thread-0  
> block here 1  
> block here 2  
> block here 3  
> finish thread-0  
> start thread-1  
> block here 1  
> block here 2

**存在问题**: 第一次启动线程,可以正常显示图片,但第二个线程就会在cv2.imshow(self.win\_name, frame)这里阻塞,不会执行下去

--------------------

## 2.解决方法 ##

opencv中imshow等显示操作只能在主线程中显示,不能在子线程中显示,那为什么第1个子线程可以正常显示咧? 原因嘛?

解决方法也很简单,既然不能在子线程中显示,但没说不能在多进程中显示呀

所有把**threading**改为**multiprocessing**,问题解决:

>     # -*-coding: utf-8 -*-
>     """
>         @Project: python-learning-notes
>         @Author : panjq
>         @E-mail : pan_jinquan@163.com
>         @Date   : 2020-04-10 18:24:06
>     """
>     
>     import os
>     import cv2
>     import glob
>     # import threading
>     import multiprocessing
>     
>     
>     class DisplayTask(object):
>         """
>         """
>     
>         def __init__(self, ):
>             """
>             :param width:
>             :param height:
>             """
>             self.win_name = "frame"
>     
>         def show(self, image_path):
>             print("block here 1")
>             frame = cv2.imread(image_path)
>             print("block here 2")
>             cv2.imshow(self.win_name, frame)
>             print("block here 3")
>             cv2.waitKey(1000)
>             cv2.destroyAllWindows()
>     
>     
>     def display_task(th_id, image_path):
>         print("start thread-{}".format(th_id))
>         dt = DisplayTask()
>         dt.show(image_path)
>     
>     
>     def demo(image_dir):
>         image_list = glob.glob(os.path.join(image_dir, "*.jpg"))
>         print("have image:{}".format(len(image_list)))
>         for th_id, image_path in enumerate(image_list):
>             # thread = threading.Thread(target=display_task, args=(th_id, image_path,))
>             thread = multiprocessing.Process(target=display_task, args=(th_id, image_path,))
>             thread.start()
>             # 等待线程执行完毕
>             thread.join()
>             print("finish thread-{}".format(th_id))
>     
>     
>     if __name__ == "__main__":
>         image_dir = "/media/dm/dm1/git/python-learning-notes/data"
>         demo(image_dir)
> 
>