PIL处理图片指定以CMYK模式

曾经终败给现在 2022-01-11 09:15 344阅读 0赞

把文件夹里，每张尺寸7200\*6800的图，按每行72个、每列68个 256\*256的图来抠出：

import os
    from PIL import Image
    import numpy as np
    
    rootdir = r'C:\Users\admin\Desktop\U_data\11/'
    savedir = r'C:\Users\admin\Desktop\U_data\22/'
    list = os.listdir(rootdir)
    j = 0
    for i in list:
        print(i)
        img = Image.open(os.path.join(rootdir, i))
        w = np.array(img).shape[0]
        h = np.array(img).shape[1]
        for x_idx in range(w // 100):
            for y_idx in range(h // 100):
                if x_idx * 100 + 256 < w and y_idx * 100 + 256 < h:
                    img1 = Image.fromarray(np.array(img)[x_idx * 100:x_idx * 100 + 256, y_idx * 100:y_idx * 100 + 256, :],mode='CMYK')
                    img1.save(os.path.join(savedir, "{}_{}_{}".format(str(x_idx), str(y_idx), i)))
                    j += 1
                    print(j)

原遥感图：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM5OTM4NjY2_size_16_color_FFFFFF_t_70][]

遥感图是4维的，如果不指定用CMYK模式的PIL读取，则截取后得到的图是这样的：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM5OTM4NjY2_size_16_color_FFFFFF_t_70 1][]

可见，颜色失真了。

而img1 = Image.fromarray(np.array(img)\[x\_idx \* 100:x\_idx \* 100 + 256, y\_idx \* 100:y\_idx \* 100 + 256, :\],mode='CMYK')后，就不会失真了：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM5OTM4NjY2_size_16_color_FFFFFF_t_70 2][]

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM5OTM4NjY2_size_16_color_FFFFFF_t_70]: /images/20220110/0e466fb254c242b188f72b2602c3e564.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM5OTM4NjY2_size_16_color_FFFFFF_t_70 1]: /images/20220110/2e43aa38bee448a4812572c640a502af.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM5OTM4NjY2_size_16_color_FFFFFF_t_70 2]: /images/20220110/c67517d4e14b49c0a5ac339e01aea2e2.png