OpenCV - matchTemplate image template matching

Template matching is the comparison of templates with overlapping image regions,An image processing method to locate coincident regions,本文记录 OpenCV Related content implementation method.

简介

Template matching tasks require searching for templates in images,A technique to determine where a template is located,Python OpenCV The function encapsulated in is cv2.matchTemplate

• 官方文档：https://docs.opencv.org/2.4.13.7/modules/imgproc/doc/object_detection.html?highlight=matchtemplate#cv2.matchTemplate

函数说明

function reference form

cv2.matchTemplate(image, templ, method[, result]) → result

参数说明

损失函数

method=CV_TM_SQDIFF

Calculate the mean squared error directly as a loss function

method=CV_TM_SQDIFF_NORMED

The mean square error loss function is calculated after normalizing the two vector modulo lengths

method=CV_TM_CCORR

Calculate the cross-correlation function result as the loss function

method=CV_TM_CCORR_NORMED

Calculate the result of the cross-correlation function normalized according to the modulo length as the loss function,个人比较推荐,效果也较好,If the image is not suitable for using this parameter directly, you can find a way to construct a normalized correlation loss function

method=CV_TM_CCOEFF

Decentralized correlation loss function

• 其中：

method=CV_TM_CCOEFF_NORMED

相关系数损失函数

示例引用

out_v = cv2.matchTemplate(image, template, cv2.TM_CCORR_NORMED)

参考资料

• https://docs.opencv.org/2.4.13.7/modules/imgproc/doc/object_detection.html?highlight=matchtemplate#cv2.matchTemplate
• https://baike.baidu.com/item/相关系数/3109424?fr=aladdin