Catalog
One 、numpy
1、 Definition
2、 library
3、ndarray The creation of
4、 function
（1）arange Section
（2）linspace A sequence of equal differences
（3）zeros whole 0 matrix
（4）ones whole 1 matrix
（5） Custom rule generation ndarray
5、 attribute
（1）dtype
（2）shape
（3）size
（4）ndim
（5）itemsize
6、 Slice indices
7、 Capture pictures
8、numpy radio broadcast
9、 Picture change
（ One ） Flip
（ Two ） tailoring
（ 3、 ... and ） The picture brightens and darkens
Two 、Pandas（ To be added ）
A very fast math library
Import and rename
import numpy as np # rename 0 Axis 1 Axis 2 Axis ...？
a=np.array([1,2,3,4],dtype=float)
print(a)
#[1. 2. 3. 4.]Parameters （ and list almost ）：（start,end,step）--- Start number , End number , Interval interval
： Floating point numbers 2.0 To 3.0 There are five numbers between , So every 0.25 Output a number
aa = np.linspace(2.0, 3.0, num=5)
print(aa)
#[2. 2.25 2.5 2.75 3. ]# whole 1 matrix
b=np.ones((3,3),np.int)
print(b,type(b))
#[[1 1 1]
#[1 1 1]
#[1 1 1]] <class 'numpy.ndarray'>
# Custom rule generation ndarray
def func(i):
return i % 4 + 1
res=np.fromfunction(func, (10,))
print(res)
#[1. 2. 3. 4. 1. 2. 3. 4. 1. 2.]Attributes of each element
# No type specified
a=np.array([1,2,3,4])
print(a)
#[1 2 3 4]
# To specify dtype=float
a=np.array([1,2,3,4],dtype=float)
print(a)
#[1. 2. 3. 4.]example 1： Read the dimension of the picture
img=cv2.imread("lyf.jpg")
print(img,type(img),img.shape)
# ...
# [126 112 113]
# [145 131 132]
# [137 123 124]]] <class 'numpy.ndarray'> (461, 438, 3)example 2： Dimension of ordinary matrix
q = np.array([[1,2,3,4],[5,6,7,8],[7,8,9,10]])
print(q.shape)
#(3, 4)
# matrix q It's three rows and four columns Element number
Array dimensions
（1） visit Elements （ and list almost （start,end,step））
# section
s=np.arange(10)
print(s) #[0 1 2 3 4 5 6 7 8 9]
print(s[...]) # All elements [0 1 2 3 4 5 6 7 8 9]
print(s[::-1]) # All elements are in reverse order [9 8 7 6 5 4 3 2 1 0]
print(s[7:1:-2]) # from 7 To 1 Print one for each two elements , No 1[7 5 3]
（2） modify Elements
# Change the index to 4、5、6、7 The elements of
s[4:8]=45,46,57,67
print(s)
#[ 0 1 2 3 45 46 57 67 8 9]
# from 4 Start to 8（ No 8） Every two elements modify one element
s[4:8:2]=45,46
print(s)
#[ 0 1 2 3 45 5 46 7 8 9]Use cascaded classifiers
（1） find haarcascade_frontalface_alt2.xml Pull the file to the project directory
（2） Create Cascading Classifier objects
detector=cv2.CascadeClassifier("haarcascade_frontalface_alt2.xml")
（3） Test pictures detectMultiScale()
faces=detector.detectMultiScale(img,1.3,3)
print(faces)
"""
It can detect all the faces in the picture , And use the face vector Save the coordinates of each face 、 size （ Use a rectangle to represent ）, The function is called by the classifier object ：
detectMultiScale
（
self,
image, # Pictures to detect
scaleFactor=None, # Zoom ratio 1.1、1.3
minNeighbors=None, # The minimum number of adjacent rectangles forming the detection target , The default value is 3 Odd number （1,3,5）
flags=None, # Is it scalable
minSize=None, # Limit the range of the obtained target area .
maxSize=None # Limit the range of the obtained target area .
）
"""（4） Intercept the face area and display
for x,y,width,height in faces:
img=img[y:y+height,x:x+width]
cv2.imshow("face",img)
cv2.waitKey(0)
cv2.destroyAllWindows()result ：
Complete code
# Capture pictures
import cv2
img=cv2.imread("lyf.jpg")
print(img,type(img),img.shape)
# Create Cascading Classifier objects
detector=cv2.CascadeClassifier("haarcascade_frontalface_alt2.xml")
# Test pictures
faces=detector.detectMultiScale(img,1.3,3)
print(faces)
# Intercept the face area
for x,y,width,height in faces:
img=img[y:y+height,x:x+width]
cv2.imshow("face",img)
cv2.waitKey(0)
cv2.destroyAllWindows()（1） About X axial symmetry （ Flip up and down ）
import cv2
img=cv2.imread("ysjx.png")
cv2.imshow("1",img)
# cv2 Upside down
img1=img[::-1]
cv2.imshow("X",img1)
cv2.waitKey(0)
cv2.destroyAllWindows()
(2) About Y axial symmetry
: Indicates that all lines ,::-1 Indicates that the column is in reverse order , So form a horizontal flip
# Flip the picture left and right : Indicates that all lines ,::-1 Indicates that the column is in reverse order , So form a horizontal flip
img2=img[:,::-1]
(3) About the origin symmetry （ Flip up, down, left and right ）
# Along the origin The ranks are in reverse order
img3=img[::-1,::-1]
All the code ：
import cv2
img=cv2.imread("ysjx.png")
cv2.imshow("1",img)
# Along x Axis Upside down ： Row in reverse order
img1=img[::-1]
# Along Y Axis Flip the picture left and right ： Indicates that all lines ,：：-1 Indicates that the column is in reverse order , So form a horizontal flip
img2=img[:,::-1]
# Channel changes , Turn grey
# img3=img[:,:,-1]
# Along the origin The ranks are in reverse order
img3=img[::-1,::-1]
cv2.imshow("X",img1)
cv2.imshow("Y",img2)
cv2.imshow("0",img3)
cv2.waitKey(0)
cv2.destroyAllWindows()(1) Intercept the lower half
Related to height --- Row number ×0.5（ Convert to int type ）
# Intercept the lower half
img4=img[int(0.5*img.shape[0]):](2) Intercept the left half
Related to width --- Number of columns ×0.5（ Convert to int type ）
img5=img[:,:int(0.5*img.shape[1])]
All the code ：
import cv2
img=cv2.imread("ysjx.png")
cv2.imshow("1",img)
# Along x Axis Upside down ： Row in reverse order
img1=img[::-1]
# Along Y Axis Flip the picture left and right ： Indicates that all lines ,：：-1 Indicates that the column is in reverse order , So form a horizontal flip
img2=img[:,::-1]
# img3=img[:,:,-1]
# Along the origin The ranks are in reverse order
img3=img[::-1,::-1]
# Intercept the lower half
img4=img[int(0.5*img.shape[0]):]
img5=img[:,:int(0.5*img.shape[1])]
# cv2.imshow("X",img1)
# cv2.imshow("Y",img2)
# cv2.imshow("0",img3)
cv2.imshow("bottom",img4)
cv2.imshow("left",img5)
cv2.waitKey(0)
cv2.destroyAllWindows()（3） Number of intercepted rows （ Draw a line every other line ）
# Extract the number of rows and intercept
img6=img[::2]
cv2.imshow("hg2",img6)
（1） The picture darkens （img*0.5 It doesn't have to be an integer , Therefore, it is necessary to deal with ）
img7=(0.5*img).astype(np.uint8)
cv2.imshow("dark",img7)
（2） The picture brightens （ You can't just *2, Because it's possible *2 Later is greater than 255, Therefore, it is necessary to deal with ）
img8=np.clip(1.5*img,a_min=0.,a_max=255.).astype(np.uint8)
cv2.imshow("bling",img8)
（6）bgr turn rgb（ Channel changes ）
The above treatment is bgr Pattern
（7） Grayscale processing （ To be added ）
