咸鱼哥学python OpenCV之图像梯度算子-Sobel Scharr Laplacian

刺骨的言语ヽ痛彻心扉 2023-02-23 12:01 36阅读 0赞

**一、Sobel算子：下减上， 右减左**  
![在这里插入图片描述][20200707100217269.png]  
**语法：**

dst = cv.Sobel(img, ddepth=, dx=, dy=, ksize=)
    # ddepth:图像深度
    # dx,dy分别表示水平和垂直方向
    # ksize是算子的大小

**获取图像：**

path = r'F:\Python_study\python_image\opencv\image\pie.png'
    img = cv.imread(path,cv.IMREAD_GRAYSCALE)

**定义一个展示函数**

def cv_show(img, name):
        cv.imshow(name, img)
        cv.waitKey(0)
        cv.destroyAllWindows()

比如，原始图像：  
![输入图像][20200707101448832.png]  
**使用X方向的计算结果：**

dst_sobel_x = cv.Sobel(img, cv.CV_64F, 1,0, ksize=3)

![在这里插入图片描述][20200707101523805.png]  
**原因解释：**  
![在这里插入图片描述][20200707101747340.png]  
为避免截断，使用绝对值方法

dst_sobel_x = cv.convertScaleAbs(dst_sobel_x)

得到结果如下：  
![在这里插入图片描述][2020070710212665.png]  
**\# 权重相加**  
sobel\_xy = cv.addWeighted(dst\_sobel\_x,0.5, dst\_sobel\_y, 0.5, 0)  
cv\_show(sobel\_xy, ‘dst\_sobel’)  
![在这里插入图片描述][20200707102808544.png]

**二、Scharr算子**  
![在这里插入图片描述][20200707103418508.png]

# Scharr算子-对结果更敏感
    dst_scharr_X = cv.Scharr(img, cv.CV_64F, 1, 0)
    dst_scharr_Y = cv.Scharr(img, cv.CV_64F, 1, 0)
    dst_scharr_X = cv.convertScaleAbs(dst_scharr_X)
    dst_scharr_Y = cv.convertScaleAbs(dst_scharr_Y)
    dst_scharr = cv.addWeighted(dst_scharr_X, 0.5, dst_scharr_Y, 0.5, 0)

**三、拉普拉斯算子**  
**\# 拉普拉斯算子-所有值相加等于0-对噪声敏感**  
![在这里插入图片描述][20200707103444123.png]

# 拉普拉斯算子-所有值相加等于0-对噪声敏感
    dst_laplacian = cv.Laplacian(img,cv.CV_64F)
    laplacian = cv.convertScaleAbs(dst_laplacian)

为对比三种之间的区别，使用lena进行比对：  
原来的lena:  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl80NTg4NTIzMg_size_16_color_FFFFFF_t_70]

res = np.hstack((sobel_xy, dst_scharr, laplacian))
    cv_show(res, 'res')

![从左往右分别是sobel算子，scharr算子，paplacian算子][sobel_scharr_paplacian]

[20200707100217269.png]: https://img-blog.csdnimg.cn/20200707100217269.png
[20200707101448832.png]: https://img-blog.csdnimg.cn/20200707101448832.png
[20200707101523805.png]: https://img-blog.csdnimg.cn/20200707101523805.png
[20200707101747340.png]: https://img-blog.csdnimg.cn/20200707101747340.png
[2020070710212665.png]: https://img-blog.csdnimg.cn/2020070710212665.png
[20200707102808544.png]: https://img-blog.csdnimg.cn/20200707102808544.png
[20200707103418508.png]: https://img-blog.csdnimg.cn/20200707103418508.png
[20200707103444123.png]: https://img-blog.csdnimg.cn/20200707103444123.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl80NTg4NTIzMg_size_16_color_FFFFFF_t_70]: https://img-blog.csdnimg.cn/20200707105130947.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl80NTg4NTIzMg==,size_16,color_FFFFFF,t_70
[sobel_scharr_paplacian]: https://img-blog.csdnimg.cn/20200707104353623.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl80NTg4NTIzMg==,size_16,color_FFFFFF,t_70