template matching class

Author: kmolLin

templatematch/findtext.jpg

@@ -0,0 +1,60 @@
+# template matching
+import cv2
+import numpy as np
+from scipy import signal
+from pprint import pprint
+from matplotlib import pyplot as plt
+
+def show_img(title, img):
+    cv2.namedWindow(f"{title}", cv2.WINDOW_NORMAL)
+    cv2.imshow(f"{title}", img)
+    cv2.waitKey(0)
+
+
+# 需要處理成透明背景 應該就可以找到對的物體進行辨識
+path1 = "t2.jpg"
+path2 = "template1.jpg"
+
+img = cv2.imread(path1, 0)
+
+template = cv2.imread(path2, 0)
+w = template.shape[1]
+h = template.shape[0]
+mask_img = np.where(template > 0, 100, 0)
+mask_img = np.float32(mask_img)
+
+cv2.imwrite("mask.jpg", mask_img)
+
+res = cv2.matchTemplate(img, template, cv2.TM_CCORR_NORMED, mask=mask_img)
+
+cv2.normalize(res, res, 0, 1, cv2.NORM_MINMAX, -1)
+
+loc = np.where(res > 0.85)
+min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)
+print(len(loc[0]))
+i = 0
+tmp_point = []
+x, y = (0, 0)
+bottom_right = (max_loc[0] + w, max_loc[1] + h)
+cv2.rectangle(img, max_loc, bottom_right, (0, 0, 0), 2)
+imm = img[max_loc[1]:bottom_right[1], max_loc[0]:bottom_right[0]]
+for pt in sorted(zip(*loc[::-1]), key=lambda s: s[0]):
+    if np.sqrt((x - pt[0]) ** 2 + (y - pt[1]) ** 2) < 500:
+        continue
+
+    x, y = pt
+    tmp_point.append(pt)
+    # print(pt, (pt[0] + w, pt[1] + h))
+    # cv2.rectangle(img, pt, (pt[0] + w, pt[1] + h), (0, 0, 0), 1)
+
+x, y = (0, 0)
+for pt in sorted(tmp_point, key=lambda s: s[1]):
+    if np.sqrt((y - pt[1]) ** 2) < 150:
+        continue
+    x, y = pt
+    print(pt)
+    # cv2.rectangle(img, pt, (pt[0] + w, pt[1] + h), (0, 0, 0), 1)
+show_img("test", img)
+cv2.imwrite("template.jpg", imm)
+# min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)
+# print(min_val, max_val, min_loc, max_loc)