opencv-python笔记
%config IPCompleter.greedy=True ----#Tab键代码自动生成
opencv-python
测试opencv导入是否成功
import cv2
img = cv2.imread("C:/Users/JD/Pictures/Saved Pictures/2.jpg", cv2.IMREAD_GRAYSCALE)
cv2.imshow("image", img)
cv2.waitKey(0)
cv2.imshow
读个视频
import cv2
img = cv2.VideoCapture("F:/pythonProject/tensflow/opencv/MP4/cxk.mp4")
# 检擦是否打开
while open:
ret, frame = img.read()
if frame is None:
break
if ret == True:
#gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
# cv2.split(frame)
# cur_img = frame.copy()
# cur_img[:, :, 2] = 0
# cur_img[:, :, 1] = 0
# cv_show("R", cur_img)
img1 = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
ret, img1 = cv2.threshold(img1, 127, 255, cv2.THRESH_BINARY)
cv2.namedWindow("result", cv2.WINDOW_NORMAL)
cv2.imshow("result", img1)
if cv2.waitKey(10) & 0XFF == 27:
break
img.release()
cv2.destroyWindow()
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
Input In [288], in | ()
21 break
22 img.release()
---> 23 cv2.destroyWindow()
TypeError: destroyWindow() missing required argument 'winname' (pos 1)
| 图像的阈值操作
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-TROQ1CFI-1673249377969)(attachment:image.png)]
图像通道的分离
# 截取部分图像数据
import cv2
img = cv2.imread("C:/Users/JD/Pictures/Saved Pictures/2.jpg")
img = img[0:500, 0:500]
def cv_show(name, img):
cv2.imshow(name, img)
cv2.waitKey(0)
cv_show("PKQ", img)
b, g, r = cv2.split(img)
img = cv2.merge((b, g, r))
# 只保留R
cur_img = img.copy()
cur_img[:, :, 0] = 0
cur_img[:, :, 1] = 0
cv_show("R", cur_img)
# 只保留G通道
cur_img = img.copy()
cur_img[:, :, 0] = 0
cur_img[:, :, 2] = 0
cv_show("G", cur_img)
# 只保留G通道
cur_img = img.copy()
cur_img[:, :, 1] = 0
cur_img[:, :, 2] = 0
cv_show("G", cur_img)
图像的平滑处理
# 滤波
# 就是卷积操作
import cv2
import numpy as np
img = cv2.imread("F:/pythonProject/tensflow/opencv/pictures/img.png")
cv2.imshow("img", img)
cv2.waitKey(0)
cv2.destroyAllWindows()
# 均值滤波
# 简单的平均卷积操作
blur = cv2.blur(img, (3, 3))
cv2.imshow('blur', blur)
cv2.waitKey(0)
cv2.destroyAllWindows()
# 方框滤波
# 基本和均值滤波一样，可以选择归一化
box = cv2.boxFilter(img, -1, (3, 3), normalize=False) # True意思是进行均一化处理
cv2.imshow('box', box)
cv2.waitKey(0)
cv2.destroyAllWindows()
# 高斯滤波
# 高斯函数：越接近于均值的时候可能性越大
# 高斯模糊的卷积核里的数值满足高斯分布，相当于更重视中间的
aussian = cv2.GaussianBlur(img, (5, 5), 1)
cv2.imshow('aussian', aussian)
cv2.waitKey(0)
cv2.destroyAllWindows()
# 中值滤波
median = cv2.medianBlur(img, 5)
cv2.imshow('median', median)
cv2.waitKey(0)
cv2.destroyAllWindows()
# 展示所有
res = np.hstack((blur, aussian, median))
print(res)
cv2.imshow("meidan vs average", res)
cv2.waitKey(0)
cv2.destroyWindow()
[[[233 218 219]
[233 218 219]
[233 218 219]
...
[233 218 219]
[233 218 219]
[233 218 219]]
[[233 218 219]
[233 218 219]
[233 218 219]
...
[233 218 219]
[233 218 219]
[233 218 219]]
[[233 218 219]
[233 218 219]
[233 218 219]
...
[233 218 219]
[233 218 219]
[233 218 219]]
...
[[236 219 220]
[236 219 220]
[236 219 220]
...
[238 220 217]
[238 220 217]
[238 220 217]]
[[236 219 220]
[236 219 220]
[236 219 220]
...
[238 220 217]
[238 220 217]
[238 220 217]]
[[236 219 220]
[236 219 220]
[236 219 220]
...
[238 220 217]
[238 220 217]
[238 220 217]]]
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
Input In [1], in | ()
44 cv2.imshow("meidan vs average", res)
45 cv2.waitKey(0)
---> 46 cv2.destroyWindow()
TypeError: destroyWindow() missing required argument 'winname' (pos 1)
|
礼帽和黑帽
- 礼帽 = 原始输入-开运算结果（留下刺）
- 黑帽 = 闭运算-原始输入
图像的梯度操作
- 梯度=膨胀-腐蚀
- sobel算子
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-F7dP03tP-1673249377970)(attachment:image.png)] - Scharr和sobel差不多，但是对变化更加的明显
- laplacian算子，对噪音点敏感
# 梯度=膨胀-腐蚀
import cv2
import numpy as np
pie = cv2.imread('F:/pythonProject/tensflow/opencv/pictures/img.png')
kernel = np.ones((7, 7), np.uint8)
dilate = cv2.dilate(pie, kernel, iterations=5)
erosion = cv2.erode(pie, kernel, iterations=5)
cv2.namedWindow("res",cv2.WINDOW_NORMAL)
res = np.hstack((dilate, erosion))
cv2.imshow("res", res)
cv2.waitKey(0)
cv2.destroyAllWindows()
gradient = cv2.morphologyEx(pie, cv2.MORPH_GRADIENT, kernel)
cv2.imshow('gradient', gradient)
cv2.waitKey(0)
cv2.destroyAllWindows()
# 计算一张图像的梯度
img = cv2.imread('F:/pythonProject/tensflow/opencv/pictures/img.png')
def cv_show(img, name):
cv2.imshow(name, img)
cv2.waitKey(0)
cv2.destroyAllWindows()
# 白到黑是正数，黑到白就是负数了，所有的附属会被截断成0，所以要取绝对值
# 分别计算x和y再求和，不建议直接求和
sobelx = cv2.Sobel(img, cv2.CV_64F, 1, 0, ksize=3)
sobelx = cv2.convertScaleAbs(sobelx)
cv_show(sobelx, "sobelx")
sobely = cv2.Sobel(img, cv2.CV_64F, 0, 1, ksize=3)
sobely = cv2.convertScaleAbs(sobely)
cv_show(sobely, "sobely")
# 求和
sobelxy = cv2.addWeighted(sobelx, 0.5, sobely, 0.5, 0)
cv_show(sobelxy, "sobelxy")
# Scharr和sobel差不多，但是对变化更加的明显
# -laplacian算子，对噪音点敏感
laplacian = cv2.Laplacian(img, cv2.CV_64F)
laplacian = cv2.convertScaleAbs(laplacian)
cv_show(laplacian, "laplacian")
Canny算子的边缘检测
# Canny边缘检测
import cv2
import numpy as np
img = cv2.imread("F:/pythonProject/tensflow/opencv/pictures/img.png")
v1 = cv2.Canny(img, 80, 150)
v2 = cv2.Canny(img, 50, 100)
res = np.hstack((v1, v2))
cv2.namedWindow("res", cv2.WINDOW_NORMAL)
cv2.imshow("res", res)
cv2.waitKey(0)
cv2.destroyAllWindows()
金字塔
- 向下采样（缩小）
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-fkj3MWlu-1673249377970)(attachment:image.png)] - 向上采样（放大）
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-YyMk8HT5-1673249377971)(attachment:image-2.png)]
import cv2
img = cv2.imread("F:/pythonProject/tensflow/opencv/pictures/img.png")
up = cv2.pyrUp(img)
cv2.imshow("up",up)
cv2.waitKey(0)
cv2.destroyAllWindows()
down = cv2.pyrDown(img)
def cv_show(name,img):
cv2.imshow("name",img)
cv2.waitKey(0)
cv2.destroyAllWindows()
cv_show(down,down)
拉普拉斯金字塔
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-j4ek4VtU-1673249377971)(attachment:image.png)]
up_down = cv2.pyrUp(down)
cv_show('up',up_down)
print(up.shape)
print(img.shape)
cv_show('img',img)
# 大小不一致能直接减
(1418, 2288, 3)
(709, 1144, 3)
轮廓的检测
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-IWsAyenG-1673249377971)(attachment:image.png)]
为了更高的准确率，使用二值图像
import numpy as np
img = cv2.imread("F:/pythonProject/tensflow/opencv/pictures/img_1.png")
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
ret, th = cv2.threshold(gray,127,255,cv2.THRESH_BINARY)
kernel = np.ones((10, 10), np.uint8)
cv2.dilate(th,kernel)
cv_show("th",th)
contours, hierarchy = cv2.findContours(th, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE)
# 传入绘制图像，轮廓，轮廓索引，颜色模式，线条厚度
# 注意需要 copy，要不原图会变。
draw_img = img.copy()
res = cv2.drawContours(draw_img, contours, -1, (255,0,0),2) #-1默认画所有的轮廓
cv_show("res",res)
轮廓特征
cnt = contours[1]
# 面积
cv2.contourArea(cnt)
---------------------------------------------------------------------------
NameError Traceback (most recent call last)
Input In [3], in | ()
----> 1 cnt = contours[1]
2 # 面积
3 cv2.contourArea(cnt)
NameError: name 'contours' is not defined
| # 计算轮廓的边长
# True 表示是闭合的
cv2.arcLength(cnt, True)
49.899494767189026
轮廓的近似
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-JLrrd0pQ-1673249377972)(attachment:image.png)]
epsilon = 0.0001*cv2.arcLength(cnt,True)
approx = cv2.approxPolyDP(cnt, epsilon,True) # 轮廓的近似
draw_img = img.copy()
res = cv2.drawContours(draw_img, approx, -1, (0,200,0),2)
cv_show('res',res)
---------------------------------------------------------------------------
NameError Traceback (most recent call last)
Input In [2], in | ()
----> 1 epsilon = 0.0001*cv2.arcLength(cnt,True)
2 approx = cv2.approxPolyDP(cnt, epsilon,True) # 轮廓的近似
4 draw_img = img.copy()
NameError: name 'cv2' is not defined
| 模板匹配
模板匹配和卷积的原理很相似，模板在原图像上开始滑动，计算模板与（图像被模板覆盖的地方）的差别度，这个差别程度的计算方法在opencv里面有六种，然后将每次计算的结果放入一个矩阵，作为结果输出，假如原图像是AxB大小，而模板是axb大小，则输出结果的矩阵是（A-a+1）x（B-b+1）
# 模板匹配
img = cv2.imread("F:/pythonProject/tensflow/opencv/pictures/img.png",0)
img_2 = cv2.imread("F:/pythonProject/tensflow/opencv/pictures/img_2.png",0)
img.shape
h,w = img_2.shape[:2]
---------------------------------------------------------------------------
NameError Traceback (most recent call last)
Input In [1], in | ()
1 # 模板匹配
----> 2 img = cv2.imread("F:/pythonProject/tensflow/opencv/pictures/img.png",0)
3 img_2 = cv2.imread("F:/pythonProject/tensflow/opencv/pictures/img_2.png",0)
4 img.shape
NameError: name 'cv2' is not defined
| img_2.shape
(325, 330)
[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-QPVpH1p5-1673249377972)(attachment:image-2.png)]
methods = ['cv2.TM_CCOEFF', 'cv2.TM_CCOEFF_NORMED', 'cv2.TM_CCORR',
'cv2.TM_CCORR_NORMED', 'cv2.TM_SQDIFF', 'cv2.TM_SQDIFF_NORMED']
res = cv2.matchTemplate(img, img_2,cv2.TM_SQDIFF)
res.shape
(385, 815)
min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)
min_val
356981760.0
max_val
1340058368.0
min_loc
(567, 0)
max_loc
(686, 146)
import matplotlib.pyplot as plt
for meth in methods:
img3 =img.copy()
# 匹配方法的真值
method = eval(meth)
print(method)
res = cv2.matchTemplate(img,img_2,method)
min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)
#如果是平方差匹配或者是是归一化平方差，匹配，取最小值
if method in [cv2.TM_SQDIFF, cv2.TM_SQDIFF_NORMED]:
top_left = min_loc
else:
top_left = max_loc
bottom_right = (top_left[0]+w,top_left[1]+h )
#画矩形
cv2.rectangle(img3, top_left,bottom_right,255,2)
plt.subplot(121),plt.imshow(res,cmap = 'gray')
plt.xticks([]),plt.yticks([])# 隐藏坐标轴
plt.subplot(122),plt.imshow(img3,cmap = 'gray')
plt.xticks([]),plt.yticks([])
plt.suptitle(meth)
plt.show()
4
5
2
3
0
1
插入一个shape的用法
img.shape[:2] 取彩色图片的长、宽。
如果img.shape[:3] 则取彩色图片的长、宽、通道。
关于img.shape[0]、[1]、[2]
img.shape[0]：图像的垂直尺寸（高度）
img.shape[1]：图像的水平尺寸（宽度）
img.shape[2]：图像的通道数
在矩阵中，[0]就表示行数，[1]则表示列数。
傅里叶变换和滤波器
傅里叶变化
低通滤波器
