自动拼图（同尺寸图片）

gaunthan Posted on Mar 22 2021

? Tool ? ? Python ?

> 将多张尺寸相同的图片拼接成一张大图。
> 
> - 自动决定行数和列数，可以使用 --ncol 选项指定列数。
> - 最后一行如果缺图，会使用全白图片填补。

## Usage

```sh
$ ../concat.py images --out='overview.png' --ncol=4
```

```sh
$ ./concat.py -h
usage: concat.py [-h] [--out OUT] [--verbose VERBOSE] [--ncol NCOL] path

concatenate images to one picture

positional arguments:
  path               path to images

optional arguments:
  -h, --help         show this help message and exit
  --out OUT          path to output
  --verbose VERBOSE  show details
  --ncol NCOL        number of columns
```

## Example

```sh
$ ls test
0.png  100.png  200.png
$ ./concat.py test --out=test.png --ncol=2
path: test
files:
         test/0.png
         test/100.png
         test/200.png
output path: test.png
output format: 2 x 2, 3 images
Concatenating...
done
```

Before:

![](https://leanote.com/api/file/getImage?fileId=605808afab644156280002e5)

After:

![](https://leanote.com/api/file/getImage?fileId=605808afab644156280002e6)

## Code

```python3
#!/usr/bin/env python3
# concat.py

import glob
import cv2 as cv
import argparse
import math
import sys

OUTPUT = 'out.png'

parser = argparse.ArgumentParser(description='concatenate images to one picture')
parser.add_argument('path', type=str, default='.', help='path to images')
parser.add_argument('--out', type=str, default=OUTPUT, help='path to output')
parser.add_argument('--verbose', type=bool, default=False, help='show details')
parser.add_argument('--ncol', type=int, default=0, help='number of columns')

args = parser.parse_args()
print('path:', args.path)

files = glob.glob(args.path + '/*.png')
print('files:')
for f in files:
    print('\t', f)

print('output path:', args.out)

if len(files) <= 1:
    print('files to concatenate should be at least 2')
    sys.exit(0)

num = len(files)
ncol = args.ncol
if ncol is None or ncol == 0:
    ncol = math.ceil(math.sqrt(num))
nrow = math.ceil(num / ncol)

print('output format: {} x {}, {} images' .format(nrow, ncol, num))

print('Concatenating...')

images = [cv.imread(f) for f in files]

row_images = []
for i in range(nrow):
    begin_index = i * ncol
    end_index = i*ncol + ncol
    if args.verbose:
        print('row {}: [{}, {})' .format(i, begin_index, end_index))

row_image = cv.hconcat(images[begin_index: min(end_index, num)])
    if args.verbose:
        print('row shape:', row_image.shape)

# pad white image to tail
    if end_index > num:
        num_to_pad = end_index - num
        if args.verbose:
            print('need to pad {} white image(s)' .format(num_to_pad))

white_image = images[begin_index] * 0 + 255
        padding = cv.hconcat([white_image for _ in range(num_to_pad)])
        if args.verbose:
            print('pad shape:', padding.shape)

row_image = cv.hconcat([row_image, padding])

if args.verbose:
            print('after padded:', row_image.shape)

row_images.append(row_image)

out = cv.vconcat(row_images)

cv.imwrite(args.out, out)

print('done')
```