Python + opencv：ValueError：設置一個序列的數組元素

我試圖編寫一個代碼，以我的第一張圖像作爲參考來更正我的圖像的V通道。但是我在嘗試將所有像素的校正值傳遞給圖像時出現此錯誤。Python + opencv：ValueError：設置一個序列的數組元素

我不知道該怎麼做。我應該如何重寫我的代碼的那部分？我嘗試了很多次，但仍然無法解決問題。

>>> 
RESTART: C:/Users/xxxx/AppData/Local/Programs/Python/Python36/avergae.py 
[0.0, 103.81328149045861, 102.25728890139274, 100.11808781708474, 102.70660218168091, 104.8367051139934, 99.823930500250071, 104.96426229104148, 101.85104381075587, 102.09709583116921, 99.400945032365726, 92.15991298604699, 101.19626441549323, 103.19529341359842, 101.34438951969196, 102.6448956449741, 94.161672541871852, 91.460941106879034, 101.18572887210487, 101.6783903260298, 90.000500755040008] 
103.81328149 
[0.0, 0.0, 1.5559925890658661, 3.6951936733738648, 1.1066793087777, -1.0234236235347964, 3.9893509902085356, -1.1509808005828717, 1.9622376797027385, 1.716185659289394, 4.4123364580928808, 11.653368504411617, 2.6170170749653749, 0.6179880768601862, 2.4688919707666486, 1.168385845484508, 9.6516089485867553, 12.352340383579573, 2.6275526183537323, 2.134891164428808] 
[[ 38 38 38 ..., 37 37 36] 
[ 38 37 38 ..., 38 38 38] 
[ 39 39 39 ..., 38 38 38] 
..., 
[141 141 142 ..., 160 161 161] 
[142 142 144 ..., 164 160 159] 
[142 142 143 ..., 168 162 159]] 
3648 
5472 
Traceback (most recent call last): 
    File "C:/Users/Lian Ming/AppData/Local/Programs/Python/Python36/avergae.py", line 49, in <module> 
    v[i,j] = v+int(deltaList[z]) 
ValueError: setting an array element with a sequence. 

>>>

代碼：

import cv2 
import numpy as np 

path = 'C:\\Users\\xxxx\\Desktop\\experiment\\aligned\\' 
path1 = 'C:\\Users\\xxxx\Desktop\\experiment\\aligned\\New folder\\' 

img1 = cv2.imread(path + 'aligned_IMG_1770.png') 
img1 = cv2.cvtColor(img1, cv2.COLOR_BGR2HSV) 
h_reference, s_reference, v_reference = cv2.split(img1) 
Average_V_Reference = np.average(v_reference) #get the average of V for my first images as a reference value to compare to the rest of images 

def ValueMean(im_file): #Create a function that do the average on the V channel for all the images 
    im = cv2.imread(im_file) 
    im = cv2.cvtColor(im, cv2.COLOR_BGR2HSV) 
    h,s,v = cv2.split(im) 
    a = np.average(v) 
    return a    

myList = [0.0] 
deltaList = [0.0] #store it in deltalist 
num_images = 20 

for i in range(1770, 1790): 
    image_name = path + 'aligned_IMG_' + str(i) + '.png' 
    myList.append(ValueMean(image_name)) #store all the images average into a list 

print(myList) 
print(Average_V_Reference) 

for z in range(1, num_images): 
    deltaList.append(z) 
    deltaList[z] = Average_V_Reference - myList[z] #Data for all the difference in average value compared to reference images 
print(deltaList) 

z=1 
for k in range(1770,1790): #create a loop to recreate images based on the data i got 
    a = 'aligned_IMG_' + str(k) 
    image_name = path + a + '.png' 
    img_file = cv2.imread(image_name) 
    img_file = cv2.cvtColor(img_file, cv2.COLOR_BGR2HSV) 
    h,s,v = cv2.split(img_file) 
    print(v) 
    print(img_file[:,:,2].shape[0]) 
    print(img_file[:,:,2].shape[1]) 
    for i in range(img_file[:,:,2].shape[0]): #passing correction value to each pixel on the V channel 
     for j in range (img_file[:,:,2].shape[1]): 
      v[i,j] = v+int(deltaList[z]) 
    z += 1 
    img_file = cv2.merge((h,s,v)) #Merge back the HSV channel 
    img_file = cv2.cvtColor(img_file, cv2.COLOR_HSV2BGR) #convert back to BGR and save 
    new_image_name = path1 + 'BrightnessHSV_%d'%i + '.png' 
    cv2.imwrite('new_image_name', new_image_name)

來源

2017-08-05 SacreD

'v [I，J] = V + INT（deltaList [Z]）'...在左邊，你有'v [i，j]' - 數組中的一個元素。在右邊，你有'v'，一個完整的數組。將整個圖像分配給單個像素不太合理。問題是，爲什麼那些循環首先出現，當一個簡單的'v + = int（deltaList [z]）'可以用更少的代碼實現同樣的事情並且更快。 –

你的意思是'v [i，j] = v [i，j] + int（deltaList [z]）' –

@DanMašek謝謝你的寶貴意見和建議......非常感謝。你知道我應該添加什麼來防止溢出或下溢嗎？ – SacreD

改變下面的行會解決它

v[i,j] = v[i,j] +int(deltaList[z])

來源

2017-08-05 17:49:32

是的。這仍然是次優解決方案。由於它將相同的值添加到矩陣中的所有元素，因此可以[一步完成]執行操作（https://docs.scipy.org/doc/numpy/reference/arrays.ndarray.html#arithmetic-matrix - 複製和比較操作）和更有效的方式。 –

Python + opencv：ValueError：設置一個序列的數組元素

回答

相關問題