12. Template matching
Template matching: In layman's terms, it is to find a picture with a picture, and find its position in the picture through a part of the picture
Execute template matching, using three matching methods
- cv.TM_SQDIFF_NORMED: Use the square difference between the two to match, the best matching value is 0;
- cv.TM_CCORR_NORMED: Use the product of the two to match, the larger the value, the better the matching degree,
- cv.TM_CCOEFF_NORMED: Match with the correlation coefficient of the two, 1 means perfect match, -1 means worst match
code show as below:
# 引入包
import cv2 as cv
import numpy as np
def template_image():
# 读取模板图片
tpl = cv.imread('./static/image/cut1.jpg')
# 读取目标图片
target = cv.imread('./static/image/windows.jpg')
cv.imshow("model", tpl)
cv.imshow("image", target)
methods = [cv.TM_SQDIFF_NORMED, cv.TM_CCORR_NORMED, cv.TM_CCOEFF_NORMED]
# 获得模板图片的高宽尺寸
th, tw = tpl.shape[:2]
for md in methods:
print(md)
# 执行模板匹配,采用的匹配方式有三种
result = cv.matchTemplate(target, tpl, md)
# 寻找矩阵(一维数组当做向量,用Mat定义)中的最大值和最小值的匹配结果及其位置
min_val, max_val, min_loc, max_loc = cv.minMaxLoc(result)
# 对于cv2.TM_SQDIFF及cv2.TM_SQDIFF_NORMED方法min_val越趋近与0匹配度越好,匹配位置取min_loc
# min_loc:矩形定点
if md == cv.TM_SQDIFF_NORMED:
tl = min_loc
# 对于其他方法max_val越趋近于1匹配度越好,匹配位置取max_loc
else:
tl = max_loc
# 矩形的宽高
br = (tl[0]+tw, tl[1]+th)
# 绘制矩形边框,将匹配区域标注出来
# tl:矩形定点
# br:矩形的宽高
# (0,0,225):矩形的边框颜色;2:矩形边框宽度
cv.rectangle(target, tl, br, (0, 0, 255), 2)
# np.str(md)
# 匹配值转换为字符串
# 显示结果,并将匹配值显示在标题栏上
cv.imshow("pipei"+np.str(md), target)
template_image()
cv.waitKey(0)
cv.destroyAllWindows()
operation result: