当我尝试将黑白PIL图像转换为numpy数组时，我遇到了奇怪的错误。我正在使用的代码示例如下。
if image.mode != '1':
image = image.convert('1') #convert to B&W
data = np.array(image) #Have also tried np.asarray(image)
n_lines = data.shape[0] #number of raster passes
line_range = range(data.shape[1])
for l in range(n_lines):
# process one horizontal line of the image
line = data[l]
for n in line_range:
if line[n] == 1:
write_line_to(xl, z+scale*n, speed) #conversion to other program code
elif line[n] == 0:
run_to(xl, z+scale*n) #conversion to other program code
我尝试使用数组和asarray进行转换，并得到了不同的错误。如果我使用数组，那么我得到的数据就像我放入的数据一样。它看起来像是几个非常萎缩的部分图像并排，图像空间的其余部分用黑色填充。如果我使用asarray，那么整个python在光栅步骤中崩溃（在随机线上）。如果我使用灰度图像（'L'），那么对于数组或asarray都不会出现这些错误。
有谁知道我做错了什么？ PIL编码B＆amp; W图像的方式有什么奇怪的，或者我需要通过numpy使其正确转换的特殊内容吗？
答案 0 :(得分：7)
我相信您在PIL中发现了一个错误！ （或者可能是笨拙的，但我会在事情的PIL方面下注......）
@ c上面的回答给出了一个解决方法（使用im.getdata（）），虽然我不确定为什么numpy.asarry（image）对他来说是segfaulting ...（旧版本的PIL和/或numpy，也许？）它适用于我，但在1位PIL图像上产生乱码（并且适用于其他所有内容，我经常使用它！）。
另一种解决方法是在转换为numpy数组之前将BW图像转换回灰度（模式“L”）。
如果速度很重要，在转换为numpy数组之前将BW图像转换回灰度似乎更快。
In [35]: %timeit np.array(im_bw.convert('L')).astype(np.uint8)
10000 loops, best of 3: 28 us per loop
In [36]: %timeit np.reshape(im_bw.getdata(), im_bw.size)
10000 loops, best of 3: 57.3 us per loop
单独注意，如果你正在修改数组内容，请务必使用numpy.array而不是numpy.asarray，因为后者将从PIL图像实例创建一个数组而不复制内存，因此返回只读数组。刚提到这个，因为我在下面使用asarray（）...
这是一个确认错误的独立示例......
import numpy as np
import Image
x = np.arange(256, dtype=np.uint8).reshape((16,16))
print 'Created array'
print x
im = Image.fromarray(x)
print 'Vales in grayscale PIL image using numpy.asarray <-- Works as expected'
print np.asarray(im)
print 'Converted to BW PIL image...'
im_bw = im.convert('1')
print 'Values in BW PIL image, using Image.getdata() <-- Works as expected'
print ' (Not a simple threshold due to dithering...)'
# Dividing by 255 to make the comparison easier
print np.reshape(im_bw.getdata(), (16, 16)) / 255
print 'Values in BW PIL image using numpy.asarray() <-- Unexpected!'
print ' (Same occurs when using numpy.array() to copy and convert.)'
print np.asarray(im_bw).astype(np.uint8)
print 'Workaround, convert back to type "L" before array conversion'
print np.array(im_bw.convert('L')).astype(np.uint8) / 255
哪个输出：
Created array
[[ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15]
[ 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31]
[ 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47]
[ 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63]
[ 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79]
[ 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95]
[ 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111]
[112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127]
[128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143]
[144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159]
[160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175]
[176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191]
[192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207]
[208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223]
[224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239]
[240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255]]
Vales in grayscale PIL image using numpy.asarray <-- Works as expected
[[ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15]
[ 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31]
[ 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47]
[ 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63]
[ 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79]
[ 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95]
[ 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111]
[112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127]
[128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143]
[144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159]
[160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175]
[176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191]
[192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207]
[208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223]
[224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239]
[240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255]]
Converted to BW PIL image...
Values in BW PIL image, using Image.getdata() <-- Works as expected
(Not a simple threshold due to dithering...)
[[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
[0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0]
[0 0 1 0 1 0 1 0 1 0 0 0 1 0 0 0]
[0 1 0 0 0 0 0 0 0 0 1 0 0 1 0 1]
[0 0 0 1 0 1 0 1 0 1 0 1 0 0 0 0]
[1 0 1 0 1 0 1 0 1 0 0 0 1 1 0 1]
[0 1 0 1 0 0 1 0 0 1 1 0 1 0 1 0]
[1 0 1 0 1 0 1 1 0 1 0 1 0 1 0 1]
[0 1 0 1 0 1 0 1 0 1 1 0 1 1 0 1]
[1 1 0 1 1 1 1 0 1 1 0 1 1 0 1 1]
[1 0 1 1 0 1 0 1 1 0 1 1 0 1 1 0]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 0 1 0 1 1 0 1 1 0 1 1 1 0 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]]
Values in BW PIL image using numpy.asarray() <-- Unexpected!
(Same occurs when using numpy.array() to copy and convert.)
[[0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[0 1 1 1 1 1 1 0 0 0 0 0 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]]
Workaround, convert back to type "L" before array conversion
[[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
[0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0]
[0 0 1 0 1 0 1 0 1 0 0 0 1 0 0 0]
[0 1 0 0 0 0 0 0 0 0 1 0 0 1 0 1]
[0 0 0 1 0 1 0 1 0 1 0 1 0 0 0 0]
[1 0 1 0 1 0 1 0 1 0 0 0 1 1 0 1]
[0 1 0 1 0 0 1 0 0 1 1 0 1 0 1 0]
[1 0 1 0 1 0 1 1 0 1 0 1 0 1 0 1]
[0 1 0 1 0 1 0 1 0 1 1 0 1 1 0 1]
[1 1 0 1 1 1 1 0 1 1 0 1 1 0 1 1]
[1 0 1 1 0 1 0 1 1 0 1 1 0 1 1 0]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 0 1 0 1 1 0 1 1 0 1 1 1 0 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]]
答案 1 :(得分：3)
不确定这一行：
data = numpy.array(image)
事实上，这给了我一个段错误。但我只是尝试了以下内容，它运行良好：
import numpy
import Image
im = Image.open("some_photo.jpg")
im = im.convert("1")
pixels = im.getdata() # returns 1D list of pixels
n = len(pixels)
data = numpy.reshape(pixels, im.size) # turn into 2D numpy array
for row in data:
# do your processing
pass
# Check that the numpy array's data is good
im2 = Image.new("1", im.size)
im2.putdata(numpy.reshape(data, [n, 1]))
im2.show()
答案 2 :(得分：0)
你的 numpy 版本是什么？我发现将 numpy 从 1.21 降级到 1.20 后，它起作用了。
df<-structure(list(serial = c(11011209, 11011209, 11011210), pnum = c(1,
2, 2), `diary day` = c(4, 4, 5), work_minutes = c(450, 450, 480
), work_hours = c(8, 8, 0), work_minutes_squared = c(230400,
230400, 0), cooking_betw = c(0.159090909, 0.371212121, 0.271428571
), dishwas_betw = c(NA, 0.28030303, 0.271428571), houseclean_betw = c(NA_real_,
NA_real_, NA_real_), laundry_betw = c(NA_real_, NA_real_, NA_real_
), ironing_betw = c(NA_real_, NA_real_, NA_real_), tv1_betw = c(0.227272727,
0.28030303, 0.088095238), tv2_betw = c(NA_real_, NA_real_, NA_real_
), tv3_betw = c(NA_real_, NA_real_, NA_real_), tv4_betw = c(0.242424242,
0.242424242, NA), tv5_betw = c(NA_real_, NA_real_, NA_real_),
tv6_betw = c(NA, NA, NA), tv7_betw = c(NA_real_, NA_real_,
NA_real_), employment = structure(c(1L, 2L, 2L), .Label = c("FT",
"PT"), class = "factor"), cooking_timedep = c(2.773774781,
2.773774781, 3.417989523), dishwas_timedep = c(1.99418589,
1.99418589, 2.625000906), houseclean_timedep = c(5.327815177,
5.327815177, 4.750929283), laundry_timedep = c(0.964285364,
0.964285364, 3.126314747), ironing_timedep = c(2.559321793,
2.559321793, 2.292682967), TV_timedep = c(5.086074482, 5.086074482,
5.012631788), TV_dur = c(70, 240, 60), Dishwash_dur = c(0,
20, 10), Foodprep_dur = c(80, 150, 10), Cleaningdwelling_dur = c(0,
0, 10), Ironing_dur = c(0, 0, 0), Laundry_dur = c(0, 0, 0
), WorkType = c("RE", "RE", "RE"), DMSex = c(1, 2, 1), DVAge = c(69,
60, 36), DVHsize = c(2, 2, 4), Income = c(1500, 1500, 3500
), NumChild = c(0, 0, 2)), row.names = c(NA, -3L), class = c("tbl_df",
"tbl", "data.frame"))