一般的变换放大缩小(都是由插值算法得到的,但是都会有损失,目前的超分辨率网络,应该是最好的图像算法,到时候会讲一下超分辨率网络)
直接进入opencv函数(具体插值算法可百度,简单的很)
例子
import cv2
image=cv2.imread("/home/dfy/Pictures/Camera_photo/Camera_photo/sss.jpg")
image1=cv2.resize(image,(1300,1200))
cv2.imshow("",image1)
cv2.waitKey(0)
if __name__ == '__main__':
print()
hist=cv2.calcHist(images="",channels="",mask="",hist="",histSize="",accumulate="")
images输入图像
channels 输入图像的通道
mask掩模图计算全部图的时候直接为None如果计算部分图像需要配相应的mask
histSize灰度级的个数
range像素值范围
例子
import cv2
from matplotlib import pyplot as plt
def main():
image = cv2.imread("/home/dfy/Pictures/Camera_photo/Camera_photo/page2.jpg")
chans=cv2.split(image)
colors=("b","g","r")
plt.figure()#创建画布
plt.title("tu")
plt.xlabel("bin")
plt.ylabel("#of pixes")
for (chan,color) in zip(chans,colors):
hist = cv2.calcHist([chan], [0], None, [256], [0, 256])
plt.plot(hist,color=color)
plt.xlim([0,256])
plt.show()
input()
if __name__ == '__main__':
main()
使用几何变换实现哈哈镜子特效
输人图像f(x, y),宽高分别为Width和Height,设置图像中心坐标Center (cx, xy)为缩放中心点,图像上任意一点到中心点的相对坐标x-Cx, ty=y- -cY。 哈哈镜效果分为图像拉伸放大和图像缩小。
对于图像拉伸放大,设置图像变换的半径为radius,哈哈镜变换后的图像为p(x, y)。
x = ( t x / 2 ) × ( s q r t ( t x × t x + t y × t y ) / r a d i u s ) + c x x= (tx/2)\times (sqrt(tx \times tx+ty \times ty)/radius) +cx x=(tx/2)×(sqrt(tx×tx+ty×ty)/radius)+cx
y = ( t y / 2 ) × ( s q r t ( t x × t x + t y × t y ) / r a d i u s ) + c y y= (ty/2) \times (sqrt(tx \times tx+ty \times ty)/radius) +cy y=(ty/2)×(sqrt(tx×tx+ty×ty)/radius)+cy
对于图像缩小,设置图像变换的半径为radius,哈哈镜变换后的图像为p(x, y)。
x = c o s ( a t a n 2 ( t y , t x ) ) 12 ( s q r t ( t x x t x + t y x t y ) + c x x= {cos(atan2(ty, tx))}{12}{(sqrt(txxtx +tyxty) +cx} x=cos(atan2(ty,tx))12(sqrt(txxtx+tyxty)+cx
y = s i n ( a t a n 2 ( t y , b x ) ) 12 ( s q r t ( t x x t x + t y x y ) + c y y= sin(atan2(ty, bx))12 (sqrt(txxtx+tyxy) +cy y=sin(atan2(ty,bx))12(sqrt(txxtx+tyxy)+cy
例子1
#可以自己调整中心点
import cv2
import math
def maxframe():
frame = cv2.imread("/home/dfy/PycharmProjects/GAN-TTS-master/sss.jpg")
height, width, n = frame.shape
center_x = width / 2
center_y = height / 2
randius = 400 # 直径
real_randius = int(randius / 2) # 半径
new_data = frame.copy()
for i in range(width):
for j in range(height):
tx = i - center_x
ty = j - center_y
distance = tx ** 2 + tx ** 2
# 为了保证选择的像素是图片上的像素
if distance < randius ** 2:
new_x = tx / 2
new_y = ty / 2
# 图片的每个像素的坐标按照原来distance 之后的distance(real_randius**2)占比放大即可
new_x = int(new_x * math.sqrt(distance) / real_randius + center_x)
new_y = int(new_y * math.sqrt(distance) / real_randius + center_y)
# 当不超过new_data 的边界时候就可赋值
if new_x < width and new_y < height:
new_data[j][i][0] = frame[new_y][new_x][0]
new_data[j][i][1] = frame[new_y][new_x][1]
new_data[j][i][2] = frame[new_y][new_x][2]
cv2.imshow("", new_data)
cv2.waitKey(0)
def MinFrame():
frame = cv2.imread("/home/dfy/PycharmProjects/GAN-TTS-master/sss.jpg")
height,width,n=frame.shape
center_x=width/2
center_y=height/2
new_data=frame.copy()
for i in range(width):
for j in range(height):
tx=i-center_x
ty=j-center_y
theta=math.atan2(ty,tx)
radius=math.sqrt(tx**2+ty**2)
new_x=int(center_x+math.sqrt(radius)*20*math.cos(theta))
new_y=int(center_y+math.sqrt(radius)*20*math.sin(theta))
if new_x<0 and new_x>width:
new_x=0
elif new_y<0 and new_y>height:
new_y=0
else:
new_data[j][i][0] = frame[new_y][new_x][0]
new_data[j][i][1] = frame[new_y][new_x][1]
new_data[j][i][2] = frame[new_y][new_x][2]
cv2.imshow("", new_data)
cv2.waitKey(0)
if __name__ == '__main__':
# maxframe()
MinFrame()