我如何在 nrrd 文件中将灰色转换为 RGB

Question

i trying convert grayscale 3D image to RGB 3D image.

Now i can get each slice's array. this array value's grayscale pixel value.

but i don't know how to covert RGB value.

i tried convert color using opencv function.

import numpy as np
import nrrd
import cv2

data = nrrd.read('C:\\Users\\admin\\Desktop\\sample data\\sample_nrrd.nrrd')
print(data[0][442][376])
cv2.cvtColor(data,cv2.COLOR_GRAY2BGR)

but it's not working...

i first time using nrrd file.

how to convert gray to rgb.

And this array is example of my data.

Thanks.

[-1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 80 236 1830 1901 1852 1742 1430 1147 1088 1285 1240 989 969 787 791 1073 1098 1380 1320 1125 1075 1209 1433 1505 1349 1114 1261 1463 1454 1696 1435 1301 1448 1384 1146 1220 1054 829 1189 1245 1319 1293 986 695 672 594 709 583 503 601 562 440 418 764 967 1275 911 842 761 652 479 691 715 505 442 768 650 705 938 1079 1076 969 936 907 902 755 588 614 770 738 646 971 802 625 890 1020 929 941 824 800 803 920 843 793 834 937 877 737 494 621 605 763 825 642 548 527 427 552 529 572 345 442 455 603 614 712 521 603 687 770 665 744 604 642 791 971 980 1059 1020 842 781 793 845 860 982 916 1077 907 491 806 533 327 709 817 913 977 735 958 624 547 651 952 1171 1184 1033 1262 2015 2193 2444 2830 2678 2650 2473 2528 2766 2915 2991 2654 2403 2700 2646 2302 2276 2706 3003 2639 2499 2414 1948 1456 1908 1409 852 500 946 747 715 864 899 960 977 807 954 1348 1053 1242 1346 1732 1634 1600 1690 1730 1797 1833 1963 1795 1775 2016 2182 2260 2132 1912 1651 1380 1576 1768 2275 1934 1790 1740 1908 2061 2068 1879 1714 1801 1678 1588 1669 1717 1596 1573 2080 1869 1922 2080 1701 2003 1617 1917 1810 1437 1292 1110 813 1079 1166 1037 1111 1518 1417 1037 603 120 137 15 -30 -197 -409 -133 -72 80 7 10 -7 -28 29 -219 -12 3 18 144 120 -89 -4 101 143 66 -162 96 218 153 120 36 188 275 58 -64 28 9 -77 89 202 206 243 349 234 54 163 262 313 282 131 175 234 102 263 109 93 57 143 282 235 175 189 217 200 297 345 314 150 -24 105 111 202 -58 20 -67 -175 -39 271 292 -2 -153 -181 -41 200 67 104 128 91 -154 -171 -42 -125 67 -172 -101 -59 -130 -94 -146 -175 23 -51 230 104 91 -16 -75 -169 -246 -203 -90 45 -99 11 72 287 149 57 111 79 -12 -104 206 0 41 68 78 -65 -255 -136 -115 53 52 61 -30 119 -155 -229 -190 -36 -163 -240 98 84 85 -17 1 54 81 -173 -205 -172 -351 -19 -86 -172 -98 -90 -169 257 126 83 171 284 297 159 50 -150 -94 -45 -39 -12 230 201 215 328 144 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -100 0 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000 -1000]

Answer 1

Your question is not very precise and accurate and there can be multiple understanding of it.

First of all there is no actual 1D color to 3D color conversion, the COLOR_GRAY2BGR only convert 1-channel grey to 3-channel grey. This conversion does not add any color, just change internal image representation so we can save it in the video . The color can not recover. Since it is a 3D depth map, I think what you want is color mapping function like the image below.

Not sure your data is the final disparity or final depth. So you have to figur out it yourself later. But the general idea is to contain it within cv::Mat and the use opencv color mapping function to colorize.

Assume data_image_1D is np array

import numpy as np 
import cv2
data_image_1D # this is the np array that you got it from somewhere
bw_img = np.reshape(data_image_1D,(rows,cols))
#rows and cols are the size of the depth image that you have. Try to see if you can get this convetsion working. 
im_color = cv2.applyColorMap(bw_img, cv2.COLORMAP_JET)
cv2.imshow("im_color",im_color)
cv2.imshow("bw_img",bw_img)
cv2.waitKey(0)
cv2.destroyAllWindows()

But still, you have to deal with the negative disparity, minimal disparity and maximum disparity issue. That one I can leave it to you

For more data mapping or custom color mapping you can follow this guide https://www.learnopencv.com/applycolormap-for-pseudocoloring-in-opencv-c-python/

edit

I think what you want might be just a simple merge function. See this

import numpy as np
import nrrd
######you orginal code put it here###
bw_img = np.reshape(data_image_1D,(rows,cols))
im_color = cv2.merge([bw_img, bw_img, bw_img])
nrrd.write(filename, im_color)# Write to a NRRD file