can' extract text exactly after crop the screen with some black blank

[linwaiwai]

can' extract text exactly after cropping the screen with some black blank or putting mutil pics in one

(x1, y1, x2, y2), image_o_shape, score, scale_infer = estimate_crop_parameters(original_file='output/embedded.png',
                                                                               template_file='output/4.png',
                                                                               scale=(0.2, 2), search_num=200)

print(f'Crop att estimate parameters: x1={x1},y1={y1},x2={x2},y2={y2}, scale_infer = {scale_infer}. score={score}')

# recover from attack:
recover_crop(template_file='output/4.png', output_file_name='output/4_a.png',
             loc=(x1, y1, x2, y2), image_o_shape=image_o_shape)

bwm1 = WaterMark(password_wm=1, password_img=1)
wm_extract = bwm1.extract('output/4_a.png', wm_shape=len_wm, mode='str')
print("截屏攻击，不知道攻击参数。提取结果：", wm_extract)
assert wm == wm_extract, '提取水印和原水印不一致'

输出

Welcome to use blind-watermark, version = 0.4.4
Make sure the version is the same when encode and decode
Your star means a lot: https://github.com/guofei9987/blind_watermark
This message only show once. To close it: `blind_watermark.bw_notes.close()`
            
Put down the length of wm_bit 215
Crop att estimate parameters: x1=165,y1=555,x2=395,y2=871, scale_infer = 0.2. score=0.2456810176372528
截屏攻击，不知道攻击参数。提取结果： 2p7�_GSx�ԟ���HuJ�jx���e
Traceback (most recent call last):
  File "/Users/linwaiwai/Documents/Project/opensource/blind_watermark/examples/example_str.py", line 79, in <module>
    assert wm == wm_extract, '提取水印和原水印不一致'
           ^^^^^^^^^^^^^^^^
AssertionError: 提取水印和原水印不一致

mutil pics in one

crop with extra black blank