[英]cv2.imwrite( ) only saves last image
I am trying to build a short script to take multiple images with my stereo-camera and save them to a directory when I press a button.我正在尝试构建一个简短的脚本,用我的立体相机拍摄多张图像,并在我按下按钮时将它们保存到一个目录中。
But for some reason I only get the last image, even if I've taken multiple pictures.但出于某种原因,即使我拍了多张照片,我也只能得到最后一张照片。 It also shows no errors and prints out the right strings as I've written in my code.它也没有显示任何错误并打印出我在代码中编写的正确字符串。 But all I get is the last image pair.但我得到的只是最后一个图像对。
I've looked at several posts but none of them have the same problem.我看了几个帖子,但没有一个有同样的问题。
EDIT: I didnt add Seconds to the image names as everyone here suggested, so the images got overwritten as they were taken during the same minute.编辑:我没有像这里的每个人建议的那样在图像名称中添加秒数,因此图像在同一分钟拍摄时被覆盖。
Here is my finished code now if anyone wants to use it:如果有人想使用它,这是我现在完成的代码:
import numpy as np
import cv2
import os
import datetime
# shape of wholeFrame: (240, 640, 3)
cap = cv2.VideoCapture(1)
now = 0
imgs_taken = 0
newpath_l = "Recorded_Images/left_imgs"
newpath_r = "Recorded_Images/right_imgs"
newpath = "Recorded_Images/both_imgs"
if not os.path.exists(newpath_l):
os.makedirs(newpath_l)
if not os.path.exists(newpath_r):
os.makedirs(newpath_r)
if not os.path.exists(newpath):
os.makedirs(newpath)
while 1:
cap.grab()
ret, wholeFrame = cap.retrieve()
if ret:
leftFrame = wholeFrame[:, 0:320, :]
rightFrame = wholeFrame[:, 320:640, :]
# Rectifying images here
stereoImage = np.concatenate((leftFrame, rightFrame), axis=1)
cv2.imshow('Take Snapshots', stereoImage)
key = cv2.waitKey(1) & 0xFF
# Saving images on keypress with timestamp
if key == ord('p'):
now = datetime.datetime.now()
if not cv2.imwrite(newpath_l + now.strftime("/img_left_%d%m%Y_%H%M%S.png"), leftFrame):
print("Left Snapshot not taken")
else:
print("Left Snapshot taken.")
if not cv2.imwrite(newpath_r + now.strftime("/img_right_%d%m%Y_%H%M%S.png"), rightFrame):
print("Right Snapshot not taken")
else:
print("Right Snapshot taken.")
if not cv2.imwrite(newpath + now.strftime("/img_both_%d%m%Y_%H%M%S.png"), stereoImage):
print("Stereo-Snapshot not taken")
else:
print("Stereo-Snapshot taken.")
imgs_taken = imgs_taken + 1
if key == ord('x'):
print("Number of images taken: " + str(imgs_taken))
break
else:
break
cv2.destroyAllWindows()
cap.release()
There is no problem with cv2.imwrite
itself, but rather in how you name the frames that you are saving. cv2.imwrite
本身没有问题,而在于您如何命名要保存的帧。 You are naming the frames as Day + Month + Year _ Hour + Minute.您将帧命名为日 + 月 + 年 _ 小时 + 分钟。 This means any frame you save within a given minute will be overridden by the last frame saved in that minute.这意味着您在给定分钟内保存的任何帧都将被该分钟内保存的最后一帧覆盖。 Saving a frame at, say, 19:00:23 will be overridden by a frame saved at 19:00:34.例如,在 19:00:23 保存的帧将被保存在 19:00:34 的帧覆盖。 Depending on your use-case, you could add + now.strftime("%S")
to be able to save one frame every second, or you could even add + now.strftime("%S_%f")
for millisecond precision.根据您的用例,您可以添加+ now.strftime("%S")
以每秒保存一帧,或者您甚至可以添加+ now.strftime("%S_%f")
以获得毫秒精度.
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.