使用`gdal`将大型彩色图像保存为`GTiff` [英] Saving a large color image as `GTiff` with `gdal`

问题描述

我正在尝试保存大小（15000,80000,3）的大图片。这个数组是一个numpy数组，我初始化为 im_final = np.zeros（（15000,80000，，3））。要进行保存，我使用 gdal ，如下所示：

I'm trying to save a large image of size (15000, 80000, 3). This array is a numpy array that I initialized as im_final = np.zeros((15000,80000,,3)). To do the saving, I use gdal like so:

dst_ds = gdal.GetDriverByName('GTiff').Create('val.tif', 80000, 15000, 3, gdal.GDT_Byte)
dst_ds.GetRasterBand(1).WriteArray(im_final[:,:,0])   # write r-band to the    raster
dst_ds.GetRasterBand(2).WriteArray(im_final[:,:,1])   # write g-band to the raster
dst_ds.GetRasterBand(3).WriteArray(im_final[:,:,2])   # write b-band to the raster
dst_ds.FlushCache()                     # write to disk
dst_ds = None

保存时，生成的图像为黑白色。但是，我需要图像为RGB，有谁知道问题是什么？此外， im_final 中的值为 uint16 。

When I save it, the resulting image is black and white. However, I need the image to be RGB, does anyone know what the issue is? Furthermore, the values in im_final are uint16.

推荐答案

问题是你试图将 uint16 写入 uint8 （ gdal.GDT_Byte ）图片。如果您确实需要一个8位图像（例如，如果您想在非GIS程序中查看此图像），最佳做法是将 im_final 缩放到0- 255。这可以是从0-65535到0-255或每个波段的最小/最大值到0-255或任何其他方式的映射。

The problem is you are trying to write uint16 into a uint8 (gdal.GDT_Byte) image. If you really need a 8-bit image (such as if you want to view this image in non-GIS programs), the best practice is to scale im_final to between 0-255. This could be a mapping from 0-65535 to 0-255 or min/max of each band to 0-255 or any other number of ways.

如果值在在 driver.Create（）。

这篇关于使用`gdal`将大型彩色图像保存为`GTiff`的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！