opencv and old games graphics python 3

Question

I am trying to match the following image

With the following templates, using cv2.matchTemplate

#elementachercher = path to template png file
#screenascanner = path to global picture where to find template


import pyautogui
import time
import cv2
import numpy as np
from matplotlib import pyplot as plt
import os
import imutils

def tryfoundobject(elementachercher, screenascanner):
    img_rgb = cv2.imread(screenascanner)
    img_gray = cv2.cvtColor(img_rgb, cv2.COLOR_BGR2GRAY)
    template = cv2.imread(elementachercher,0)
    w, h = template.shape[::-1]
    res = cv2.matchTemplate(img_gray,template,cv2.TM_CCOEFF_NORMED)
    threshold = 0.8
    loc = np.where( res >= threshold)
    for pt in zip(*loc[::-1]):
        cv2.rectangle(img_rgb, pt, (pt[0] + w, pt[1] + h), (0,0,255), 2)
        if pt != None:
            print(elementachercher+" matched")
            break
        else:
            continue
    cv2.imwrite(elementachercher,img_rgb)
    
  tryfoundobject("/home/noway/template.png", "/home/noway/mypicture.png")

But no matches are found.

What am I doing wrong? 1 screen is what result i want, the second screen is what i got after script

i m trying to use matchtemplate on the big picture, the small picture are my template for get template coordinates, X, Y of them but opencv never match my template on the main picture — VeilleurTryToFix, May 23 '21 at 09:18
what *does* happen? edit the question please to show `res` (normalize to [0,255] first), then show `loc`. — Gulzar, May 23 '21 at 09:21
i edited my post for show what im trying to do with screen, what i want, and what i get — VeilleurTryToFix, May 23 '21 at 10:18
no indentation error, my script works fine but result is not what im trying to do maybe i make a mistake during my workflow with opencv — VeilleurTryToFix, May 23 '21 at 10:25
what does happen? edit the question please to show res (normalize to [0,255] first), then show loc — Gulzar, May 23 '21 at 10:25
` normalizedImg = cv2.normalize(img_rgb, res, 0, 255, cv2.NORM_MINMAX) cv2.imshow('dst_rt', normalizedImg) cv2.waitKey(0) cv2.destroyAllWindows()` nothing change, i got exacly same picture as the last picture in my post, — VeilleurTryToFix, May 23 '21 at 10:33
Let us [continue this discussion in chat](https://chat.stackoverflow.com/rooms/232766/discussion-between-veilleurtrytofix-and-gulzar). — VeilleurTryToFix, May 23 '21 at 10:42

score 0 · Answer 1 · answered May 23 '21 at 11:07

Following the chat discussion

res_normed = (res - np.min(res)) / (np.max(res) - np.min(res))
res_uint = np.array(res_normed * 255, dtype=np.uint8)
cv2.imshow("res_uint ", res_uint)

res looks like this

You have to look for local maxima on this image to get the coordinates of matches.

Refer to this for a very simple POC and to this for the full solution.

opencv and old games graphics python 3

1 Answers1