解决OpenCV代码TypeError: only integer scalar arrays can be converted to a scalar index

错误根因 #

代码存在3个问题共同触发报错：

1. 变量名冲突：开头定义块大小的变量B=8，后续拆分LAB通道时执行L,A,B=cv2.split(Trans_Color)，直接将B覆盖为了B通道的二维numpy数组。后续循环中执行row*B时，实际是整数乘数组得到数组，切片时传入数组作为索引直接触发类型错误。
2. 数值类型不匹配：计算分块数量blocksV、blocksH时使用Python3普通除法/得到的是浮点数，range()函数不支持接收浮点数作为参数。
3. 隐藏问题：计算了对齐到8像素的图像高宽h、w，但未对原始图像做裁剪，若原图高宽不是8的整数倍，后续分块处理会出现尺寸不匹配问题。

修复方案 #

调整要点 #

• 将块大小的变量名从B改为BLOCK_SIZE，避免和LAB的B通道变量冲突
• 计算完h、w后裁剪原始图像，保证后续处理的图像尺寸都是8的整数倍
• 分块数量计算改用整数除法//，保证blocksV、blocksH为整数类型

修复后完整可运行代码如下：

import cv2
import numpy as np

BLOCK_SIZE=8 # 块大小变量名修改，避免和B通道冲突
img1 = cv2.imread(r'C:\Users\Leith\Desktop\Test Images\Goldhill_Final.bmp')
h,w=np.array(img1.shape[:2])/BLOCK_SIZE * BLOCK_SIZE
h=int(h)
w=int(w)
# 新增：裁剪图像到对齐后的尺寸
img1 = img1[:h, :w]

Trans_Color=cv2.cvtColor(img1, cv2.COLOR_BGR2LAB)
Trans_Color=Trans_Color.astype(int)
L,A,B=cv2.split(Trans_Color)

Three_Channels=[Trans_Color[:,:,0],A,B]

QY=np.array([[16,11,10,16,24,40,51,61],
                         [12,12,14,19,26,48,60,55],
                         [14,13,16,24,40,57,69,56],
                         [14,17,22,29,51,87,80,62],
                         [18,22,37,56,68,109,103,77],
                         [24,35,55,64,81,104,113,92],
                         [49,64,78,87,103,121,120,101],
                         [72,92,95,98,112,100,103,99]])

QC=np.array([[17,18,24,47,99,99,99,99],
                         [18,21,26,66,99,99,99,99],
                         [24,26,56,99,99,99,99,99],
                         [47,66,99,99,99,99,99,99],
                         [99,99,99,99,99,99,99,99],
                         [99,99,99,99,99,99,99,99],
                         [99,99,99,99,99,99,99,99],
                         [99,99,99,99,99,99,99,99]])

QF=99.0
if QF < 50 and QF > 1:
    scale = np.floor(5000/QF)
elif QF < 100:
    scale = 200-2*QF
else:
    print("Quality Factor must be in the range [1..99]")
    
scale=scale/100.0
Q=[QY*scale,QC*scale,QC*scale]

TransAll=[]
TransAllQuant=[]
for idx,channel in enumerate(Three_Channels):
    channelrows=channel.shape[0]
    channelcols=channel.shape[1]
    Trans = np.zeros((channelrows,channelcols), np.float32)
    TransQuant = np.zeros((channelrows,channelcols), np.float32)
    # 改用整数除法得到整数类型的分块数量
    blocksV=channelrows//BLOCK_SIZE
    blocksH=channelcols//BLOCK_SIZE
    vis0 = np.zeros((channelrows,channelcols), np.float32)
    vis0[:channelrows, :channelcols] = channel
    vis0=vis0-128
    for row in range(blocksV):
        for col in range(blocksH):
            # 块大小改用修改后的BLOCK_SIZE变量
            currentblock = cv2.dct(vis0[row*BLOCK_SIZE:(row+1)*BLOCK_SIZE,col*BLOCK_SIZE:(col+1)*BLOCK_SIZE])
            Trans[row*BLOCK_SIZE:(row+1)*BLOCK_SIZE,col*BLOCK_SIZE:(col+1)*BLOCK_SIZE]=currentblock
            TransQuant[row*BLOCK_SIZE:(row+1)*BLOCK_SIZE,col*BLOCK_SIZE:(col+1)*BLOCK_SIZE]=np.round(currentblock/Q[idx])
    TransAll.append(Trans)
    TransAllQuant.append(TransQuant)

内容的提问来源于stack exchange，提问作者Leith Abed