咸鱼哥学python OpenCV之模板匹配

﹏ヽ暗。殇╰゛Y 2023-02-23 14:28 20阅读 0赞

模板匹配和卷积的原理很像， 模板在原图像上从原点开始滑动，计算模板与图像被模板覆盖的地方的差异程度，这个差异程度的计算方法有6种（后面将会提到），然后每次将计算的结果放入一个矩阵里，作为结果输出。假设原图像是AXB大小，模板是axb大小，则输出的矩阵是（A-a+1）x（B-b+1）  
**模板匹配差异度计算方法：**  
![在这里插入图片描述][20200707210619806.png]  
**单目标匹配**

import numpy as np
    import matplotlib.pyplot as plt
    import cv2 as cv
    
    img_path = r'F:\Python_study\python_image\opencv\image\lena.png'
    template_path = r'F:\Python_study\python_image\opencv\image\lena_eyes.png'
    
    img = cv.imread(img_path, 0)
    template = cv.imread(template_path, 0)
    h,w = template.shape[:2]
    print(img.shape, template.shape)
    
    methods = ['cv.TM_CCOEFF', 'cv.TM_CCOEFF_NORMED', 'cv.TM_CCORR',
                'cv.TM_CCORR_NORMED', 'cv.TM_SQDIFF', 'cv.TM_SQDIFF_NORMED']
    
    for meth in methods:
        draw_img = img.copy()
        # eval存储在字符串中的有效python表达式
        method = eval(meth)
        # 模板匹配
        res = cv.matchTemplate(img, template, method)
        print(res.shape)
        # 寻找最值
        min_val, max_val, min_loc, max_loc = cv.minMaxLoc(res)
        # 使用不同的比较方法，对结果的解释不同
    
        if method in [cv.TM_SQDIFF, cv.TM_SQDIFF_NORMED]:
            top_left = min_loc
        else:
            top_left = max_loc
        bottom_right = (top_left[0] + w, top_left[1] + h)
        cv.rectangle(draw_img, top_left, bottom_right, 255, 2)
        plt.subplot(121), plt.imshow(res, cmap='gray')
        plt.title('Matching Result'), plt.xticks([]), plt.yticks([])
        plt.subplot(122), plt.imshow(draw_img, cmap='gray')
        plt.title('Detected Point'), plt.xticks([]), plt.yticks([])
        plt.suptitle(meth)
        plt.show()

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl80NTg4NTIzMg_size_16_color_FFFFFF_t_70]  
可见归一化之后的效果更好

多目标匹配

import matplotlib.pyplot as plt
    import cv2 as cv
    img_path = r'F:\Python_study\python_image\opencv\image\chessboard.png'
    template_path = r'F:\Python_study\python_image\opencv\image\chessboard_01.png'
    img_rgb = cv.imread(img_path)
    img_gray = cv.cvtColor(img_rgb, cv.COLOR_BGR2GRAY)
    template = cv.imread(template_path,0)
    w, h = template.shape[::-1]
    res = cv.matchTemplate(img_gray,template,cv.TM_CCOEFF_NORMED)
    threshold = 0.4
    loc = np.where( res >= threshold)
    for pt in zip(*loc[::-1]):
        cv.rectangle(img_rgb, pt, (pt[0] + w, pt[1] + h), (0,0,255), 1)
    cv.imshow("img",img_rgb)

[20200707210619806.png]: https://img-blog.csdnimg.cn/20200707210619806.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl80NTg4NTIzMg_size_16_color_FFFFFF_t_70]: https://img-blog.csdnimg.cn/20200707213452136.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl80NTg4NTIzMg==,size_16,color_FFFFFF,t_70