In order to compare the effects before and after image processing , Especially the rendering before and after the algorithm processing , We need to display multiple images at the same time . Here the opencv Image mosaic method to achieve the desired effect .

1 Defined function show_multi_img

         Define picture display functions show_multi_img, Including 5 Parameters , The meanings and types of each parameter are as follows ：

        （1）scale：float type , Image scaling , That is to scale the image .

        （2）imglist：list type , That is, a list of image data to be displayed .

        （3）order：list or tuple type , Refers to the rows and columns of the image display window ,order[0] Said line ,order[1] The column . The default value is 1 That's ok N Column ,N by imglist The length of , That is, the number of images to be displayed .

        （4）border：int type , That is, the minimum interval between images .

        （5）border：tuple type , The color of the image space .

         This function is compatible with displaying pictures of different sizes at the same time .

2 Sample code

# -*- coding: utf-8 -*-
"""
 The official account of Lele perception school 
@author: https://blog.csdn.net/suiyingy
"""

import cv2
import numpy as np

def show_multi_imgs(scale, imglist, order=None, border=10, border_color=(255, 255, 0)):
    """
    :param scale: float  Scale of original image scaling 
    :param imglist: list  Image sequence to be displayed 
    :param order: list or tuple  According to the order   That's ok × Column 
    :param border: int  Image spacing distance 
    :param border_color: tuple  Bay area color 
    :return:  Back to the spliced numpy Array 
    """
    if order is None:
        order = [1, len(imglist)]
    allimgs = imglist.copy()
    ws , hs = [], []
    for i, img in enumerate(allimgs):
        if np.ndim(img) == 2:
            allimgs[i] = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
        allimgs[i] = cv2.resize(img, dsize=(0, 0), fx=scale, fy=scale)
        ws.append(allimgs[i].shape[1])
        hs.append(allimgs[i].shape[0])
    w = max(ws)
    h = max(hs)
    #  Splice the pictures to be displayed 
    sub = int(order[0] * order[1] - len(imglist))
    #  Judge the size relationship between the input display format and the number of images to be displayed 
    if sub > 0:
        for s in range(sub):
            allimgs.append(np.zeros_like(allimgs[0]))
    elif sub < 0:
        allimgs = allimgs[:sub]
    imgblank = np.zeros(((h+border) * order[0], (w+border) * order[1], 3)) + border_color
    imgblank = imgblank.astype(np.uint8)
    for i in range(order[0]):
        for j in range(order[1]):
            imgblank[(i * h + i*border):((i + 1) * h+i*border), (j * w + j*border):((j + 1) * w + j*border), :] = allimgs[i * order[1] + j]
    return imgblank


if __name__ == '__main__':
    image = cv2.imread('lena.jpg')
    img = show_multi_imgs(0.9, [image, image, image, image, image, image], (2, 3))
    cv2.namedWindow('multi', 0)
    cv2.imshow('multi', img)
    cv2.waitKey(0)

3 The test results

         The test results are shown in the following figure ：

More 3D 、 Please pay attention to two-dimensional perception algorithm and financial quantitative analysis algorithm “ Lele perception school ” WeChat official account , And will continue to update .