cv2.findContours、cv2.threshold函数的用法举例

我就是我 2023-02-11 04:52 30阅读 0赞

本文需要参考的内容：

*  cv2.threshold 请参考：[https://blog.csdn.net/weixin\_42296411/article/details/80901080][https_blog.csdn.net_weixin_42296411_article_details_80901080]
 *  cv2.findContours 请参考：[https://blog.csdn.net/Easen\_Yu/article/details/89365497][https_blog.csdn.net_Easen_Yu_article_details_89365497]

具体内容如下：

### 图片内容 ###

![在这里插入图片描述][20200521231319621.png]

### step1 ###

import cv2
    import numpy as np
    
    img = cv2.imread('binary_img.jpeg',0)  # 0代表以灰度模式读入图片
    print('img\'s shape is: {}'.format(img.shape))
    cv2.imshow('raw',img)
    cv2.waitKey(0)
    cv2.destroyAllWindows()

显示结果如下：

img's shape is: (158, 230)

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1ODY2NzM2_size_16_color_FFFFFF_t_70]

### step 2 ###

retval, dst = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY) # retval就是返回的阈值; 我们设定的阈值就是127
    print('retval is: {}'.format(retval))
    for instanceId in np.unique(dst):
        print('instanceId is: {}'.format(instanceId))
        
    cv2.imshow('dst',dst)
    cv2.waitKey(0)
    # if k ==27: # 键盘上Esc键的键值
    cv2.destroyAllWindows()

结果如下：

retval is: 127.0
    instanceId is: 0
    instanceId is: 255

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1ODY2NzM2_size_16_color_FFFFFF_t_70 1]

### step 3 ###

# mode = cv2.RETR_EXTERNAL method = cv2.CHAIN_APPROX_NONE
    contour, _ = cv2.findContours(dst.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE) 
    print('The type of contour is: {}'.format(type(contour)))
    print('{} polygon(s) in the contour'.format(np.array(contour).shape[0]))
    for i in contour:
        print(i.shape)

结果如下：

The type of contour is: <class 'list'>
    3 polygon(s) in the contour
    (40, 1, 2)
    (66, 1, 2)
    (28, 1, 2)

### step 4 ###

# reshape(-1)把数据转换成一维了
    polygons = [c.reshape(-1).tolist() for c in contour]
    for cont in polygons:
        print(np.array(cont).shape)

结果如下：

(80,)
    (132,)
    (56,)

### step 5 ###

# 通过polygon找box
    def polys_to_boxes(polys):
        """Convert a list of polygons into an array of tight bounding boxes."""
        boxes_from_polys = np.zeros((len(polys), 4), dtype=np.float32)
        for i in range(len(polys)):  # poly的shape应该是(c,n),c代表多边形的个数，n代表该多边形顶点个数的2倍。
            poly = polys[i]    # 这种写法对吗？？？
            x0 = min(poly[::2])
            x1 = max(poly[::2])
            y0 = min(poly[1::2])
            y1 = max(poly[1::2])
            boxes_from_polys[i, :] = [x0, y0, x1, y1]
        return boxes_from_polys
    
    adc = polys_to_boxes(polygons)
    print(adc)

[[158.  88. 173. 101.]
     [ 56.  64.  78.  88.]
     [143.  62. 154.  70.]]

[https_blog.csdn.net_weixin_42296411_article_details_80901080]: https://blog.csdn.net/weixin_42296411/article/details/80901080
[https_blog.csdn.net_Easen_Yu_article_details_89365497]: https://blog.csdn.net/Easen_Yu/article/details/89365497
[20200521231319621.png]: https://img-blog.csdnimg.cn/20200521231319621.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1ODY2NzM2_size_16_color_FFFFFF_t_70]: https://img-blog.csdnimg.cn/20200521224851939.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1ODY2NzM2,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1ODY2NzM2_size_16_color_FFFFFF_t_70 1]: https://img-blog.csdnimg.cn/20200521225510310.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM1ODY2NzM2,size_16,color_FFFFFF,t_70