OpenCV—python Max-Mix filter 最大-最小滤波器
Posted SongpingWang
tags:
篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了OpenCV—python Max-Mix filter 最大-最小滤波器相关的知识,希望对你有一定的参考价值。
文章目录
一、Max-Min Filter 算法原理
图像的细节属于低频信息,图像的边缘属于高频信息。我们使用一定大小的 Max-Min 滤波器作用于图像,当滤波器作用于图像细节时,输出结果往往趋向于0(黑色);而滤波器作用于图像边缘时,Max-Min 输出结果往往趋向于255(白色)。所以 最大-最小滤波器 能有效地用于检测图像的边缘和轮廓。
MAX-MIN滤波器使用网格内像素的最大值和最小值的差值对网格内像素重新赋值。通常用于边缘检测。
边缘检测用于检测图像中的线。像这样提取图像中的信息的操作被称为特征提取。边缘检测通常在灰度图像上进行。
滤波流程图:
二、代码实现:
import cv2
import numpy as np
def max_min_filter(img, K_size):
height, width = img.shape
pad = K_size // 2
out_img = img.copy()
pad_img = np.zeros((height + pad*2, width + pad*2), dtype=np.uint8)
pad_img[pad: pad+height, pad: pad+width] = img.copy()
for y in range(height):
for x in range(width):
out_img[y,x] = np.max(pad_img[y:y+K_size, x:x+K_size]) - np.min(pad_img[y:y+K_size, x:x+K_size])
return out_img
if __name__ == __main__:
gray = cv2.imread("./Test_image2.jpg", 0)
out_0 = max_min_filter(gray, 3)
out_1 = cv2.bitwise_not(out_0)
cv2.imwrite("out_0.png", out_0)
cv2.imwrite("out_1.png", out_1)
示例:阴影去除
import time
import cv2
import numpy as np
import matplotlib.pyplot as plt
def max_filtering(height,width,ksize, I_temp):
n = (ksize//2)
wall = np.zeros((height+n*2, width+n*2),dtype=np.int32)
temp = wall.copy()
wall[n:height+n, n:width+n] = I_temp.copy()
for y in range(height):
for x in range(width):
temp[y+n,x+n] = np.max(wall[y:y+ksize,x:x+ksize])
return temp[n:height+n, n:width+n]
def min_filtering(height,width,ksize, A):
n = (ksize // 2)
wall_min = np.full((height+n*2, width+n*2), 255,dtype=np.int32)
temp_min = wall_min.copy()
wall_min[n:height+n, n:width+n] = A.copy()
for y in range(height):
for x in range(width):
temp_min[y+n,x+n] = np.min(wall_min[y:y+ksize,x:x+ksize])
return temp_min[n:height+n, n:width+n]
def background_subtraction(I, B):
O = I - B
norm_img = cv2.normalize(O, None, 0,255, norm_type=cv2.NORM_MINMAX)
return norm_img
def min_max_filtering(M, ksize, I):
height,width = I.shape[:2]
if M == 0:
A = max_filtering(height,width,ksize, I)
B = min_filtering(height,width,ksize, A)
normalised_img = background_subtraction(I, B)
elif M == 1:
A = min_filtering(height,width,ksize, I)
B = max_filtering(height,width,ksize, A)
normalised_img = background_subtraction(I, B)
else:
normalised_img = None
return normalised_img
if __name__ == __main__:
Image = cv2.imread(Test_image3.jpg,0)
t0 = time.time()
output1 = min_max_filtering(0, 21, Image)
t1 = time.time()
print("Total cost time is:",t1-t0)
cv2.imwrite("Test_image3_out.png",output1)
效果图有点不干净
以上是关于OpenCV—python Max-Mix filter 最大-最小滤波器的主要内容,如果未能解决你的问题,请参考以下文章
在open cv2 python中使用Tensor flow 2.0对象的方法是啥,为啥这么迂回?
python 来自https://stackoverflow.com/questions/19677109/script-for-automatically-loading-reference-fil
Qt-OpenCV学习笔记--读取视频--VideoCapture()
opencv4opencv视频教程 C++(opencv教程)3矩阵的掩膜操作(filter2D)
OpenCV(4.4.0) 错误: (-215:Assertion failed) isMap() in function 'cv::FileNode::operator []'