一、图像增强（该代码有问题）
import os
import cv2
import random
import numpy as np
def random_crop(img, area_ratio, hw_vari):
"""
:param img:
:param area_ratio: 裁剪画面占原画面比例(0,0.5)
:param hw_vari: 扰动占原高宽比的范围(0.2)
x0: 左上角横坐标
y0: 左上角纵坐标
w: 裁剪宽度
h: 裁剪高度
:return:
"""
h, w = img.shape[:2]
hw_delta = np.random.uniform(-hw_vari, hw_vari)
hw_mult = 1 + hw_delta
w_crop = int(round(w*np.sqrt(area_ratio*hw_mult)))
if w_crop > w - 2:
w_crop = w - 2
h_crop = int(round(h*np.sqrt(area_ratio/hw_mult)))
if h_crop > h - 2:
h_crop = h - 2
x0 = np.random.randint(0, w-w_crop-1)
y0 = np.random.randint(0, h-h_crop-1)
crop_image = lambda img, x0, y0, w, h: img[y0:y0 + h, x0:x0 + w]
return crop_image
def random_rotate(img, angle_vari, p_crop):
"""
:param img:
:param angle_vari: 旋转角度（逆时针）180
:param p_crop: 去黑边裁剪比例0.5
:return:
"""
angle = np.random.uniform(-angle_vari, angle_vari)
crop = False if np.random.random() > p_crop else True
h, w = img.shape[:2]
angle %= 360
M_rotate = cv2.getRotationMatrix2D((w / 2, h / 2), angle, 1)
img_rotated = cv2.warpAffine(img, M_rotate, (w, h))
if crop:
angle_crop = angle % 180
if angle_crop > 90:
angle_crop = 180 - angle_crop
theta = angle_crop * np.pi / 180.0
hw_ratio = float(h) / float(w)
tan_theta = np.tan(theta)
numerator = np.cos(theta) + np.sin(theta) * tan_theta
r = hw_ratio if h > w else 1 / hw_ratio
denominator = r * tan_theta + 1
crop_mult = numerator / denominator
w_crop = int(round(crop_mult * w))
h_crop = int(round(crop_mult * h))
x0 = int((w - w_crop) / 2)
y0 = int((h - h_crop) / 2)
img_rotated = lambda img, x0, y0, w, h: img[y0:y0 + h, x0:x0 + w]
return img_rotated
def random_hsv_transform(img, hue_vari, sat_vari, val_vari):
"""
:param img:
:param hue_vari: 色调变化比例范围
:param sat_vari: 饱和度变化比例范围
:param val_vari: 明度变化比例范围
:return:
"""
hue_delta = np.random.randint(-hue_vari, hue_vari)
sat_mult = 1 + np.random.uniform(-sat_vari, sat_vari)
val_mult = 1 + np.random.uniform(-val_vari, val_vari)
img_hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV).astype(np.float)
img_hsv[:, :, 0] = (img_hsv[:, :, 0] + hue_delta) % 180
img_hsv[:, :, 1] *= sat_mult
img_hsv[:, :, 2] *= val_mult
img_hsv[img_hsv > 255] = 255
return cv2.cvtColor(np.round(img_hsv).astype(np.uint8), cv2.COLOR_HSV2BGR)
def random_gamma_transform(img, gamma_vari):
log_gamma_vari = np.log(gamma_vari)
alpha = np.random.uniform(-log_gamma_vari, log_gamma_vari)
gamma = np.exp(alpha)
gamma_table = [np.power(x / 255.0, gamma) * 255.0 for x in range(256)]
gamma_table = np.round(np.array(gamma_table)).astype(np.uint8)
return cv2.LUT(img, gamma_table)
def random_flip_img(img):
"""
0 = X axis, 1 = Y axis, -1 = both
:param img:
:return:
"""
flip_val = [0,1,-1]
random_flip_val = random.choice(flip_val)
res = cv2.flip(img, random_flip_val)
return res
if __name__ == '__main__':
output_dir = "./temp_dir"
input_dir = "./temp_dir"
index = 1
for child_dir in os.listdir(input_dir):
child_path = os.path.join(input_dir, child_dir)
for dir_image in os.listdir(child_path):
print(child_path)
Suffix_name = ['.png', '.jpg', '.jpeg']
if dir_image.endswith(tuple(Suffix_name)):
img = cv2.imread(os.path.join(child_path, dir_image))
random_crop_img = random_crop(img,0.2,0.2)
cv2.imwrite(child_path + '/crop_img' + str(index) + '.png', random_crop_img)
res_rotate = random_rotate(img, 180, 0.5)
cv2.imwrite(child_path + '/rotate_img' + str(index) + '.png', res_rotate)
HSV_IMG = random_hsv_transform(img,2,2,3)
cv2.imwrite(child_path + '/HSV_IMG' + str(index) + '.png', HSV_IMG)
GAMMA_IMG = random_gamma_transform(img,0.2)
cv2.imwrite(child_path + '/GAMMA_IMG' + str(index) + '.png', GAMMA_IMG)
res_flip = random_flip_img(img)
cv2.imwrite(child_path + './flip_img'+ str(index) + '.png', res_flip)
index+=1
二、图像增强
获取目录下所有文件夹以及其里面的图片文件并对其执行图像增强
import os
import cv2
import random
import numpy as np
from scipy import *
def Affine_transformation(img):
"""
:param img:
:return:
"""
h, w = img.shape[:2]
points1 = np.float32([[50, 50], [200, 50], [50, 200]])
points2 = np.float32([[30, 80], [200, 50], [80, 220]])
matrix = cv2.getAffineTransform(points1, points2)
Affine_transfor_img = cv2.warpAffine(img, matrix, (h, w))
return Affine_transfor_img
def random_rotate_img(img, min_angle, max_angle):
'''
random rotation an image
:param img: image to be rotated
:param min_angle: min angle to rotate
:param max_angle: max angle to rotate
:return: image after random rotated
'''
if not isinstance(img, list):
img = [img]
angle = random.randint(min_angle, max_angle)
center = (img[0].shape[0] / 2, img[0].shape[1] / 2)
rot_matrix = cv2.getRotationMatrix2D(center, angle, scale=1.0)
res = []
for img_inst in img:
img_inst = cv2.warpAffine(img_inst, rot_matrix, dsize=img_inst.shape[:2], borderMode=cv2.BORDER_CONSTANT)
res.append(img_inst)
if len(res) == 0:
res = res[0]
return res
def random_hsv_transform(img, hue_vari, sat_vari, val_vari):
"""
:param img:
:param hue_vari: 色调变化比例范围(0,360)
:param sat_vari: 饱和度变化比例范围(0,1)
:param val_vari: 明度变化比例范围(0,1)
:return:
"""
hue_delta = np.random.randint(-hue_vari, hue_vari)
sat_mult = 1 + np.random.uniform(-sat_vari, sat_vari)
val_mult = 1 + np.random.uniform(-val_vari, val_vari)
img_hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV).astype(np.float)
img_hsv[:, :, 0] = (img_hsv[:, :, 0] + hue_delta) % 180
img_hsv[:, :, 1] *= sat_mult
img_hsv[:, :, 2] *= val_mult
img_hsv[img_hsv > 255] = 255
return cv2.cvtColor(np.round(img_hsv).astype(np.uint8), cv2.COLOR_HSV2BGR)
def random_gamma_transform(img, gamma_vari):
"""
:param img:
:param gamma_vari:
:return:
"""
log_gamma_vari = np.log(gamma_vari)
alpha = np.random.uniform(-log_gamma_vari, log_gamma_vari)
gamma = np.exp(alpha)
gamma_table = [np.power(x / 255.0, gamma) * 255.0 for x in range(256)]
gamma_table = np.round(np.array(gamma_table)).astype(np.uint8)
return cv2.LUT(img, gamma_table)
def random_flip_img(img):
"""
0 = X axis, 1 = Y axis, -1 = both
:param img:
:return:
"""
flip_val = [0,1,-1]
random_flip_val = random.choice(flip_val)
res = cv2.flip(img, random_flip_val)
return res
def SaltAndPepper(img): #定义添加椒盐噪声的函数
"""
:param img:
:return:
"""
SP_NoiseImg=img
percetage = np.random.randint(3) * 0.1
SP_NoiseNum=int(percetage*img.shape[0]*img.shape[1])
for i in range(SP_NoiseNum):
randX=random.random_integers(0,img.shape[0]-1)
randY=random.random_integers(0,img.shape[1]-1)
if random.random_integers(0,1)==0:
SP_NoiseImg[randX,randY]=0
else:
SP_NoiseImg[randX,randY]=255
return SP_NoiseImg
def clamp(pv): #防止像素溢出
if pv > 255:
return 255
if pv < 0:
return 0
else:
return pv
def gaussian_noise(image): # 加高斯噪声
"""
:param image:
:return:
"""
h, w, c = image.shape
for row in range(h):
for col in range(w):
s = np.random.normal(0, 20, 3)
b = image[row, col, 0] # blue
g = image[row, col, 1] # green
r = image[row, col, 2] # red
image[row, col, 0] = clamp(b + s[0])
image[row, col, 1] = clamp(g + s[1])
image[row, col, 2] = clamp(r + s[2])
return image
if __name__ == '__main__':
input_dir = "./face_dataset/test_faces"
index = 1
for child_dir in os.listdir(input_dir):
child_path = input_dir +"/"+ child_dir
for dir_image in os.listdir(child_path):
Suffix_name = ['.png', '.jpg', '.jpeg']
if dir_image.endswith(tuple(Suffix_name)):
img = cv2.imread(child_path +"/"+ dir_image)
print("正在处理：",child_path +"/"+ dir_image,"...")
Affine_transfor_img = Affine_transformation(img)
cv2.imwrite(child_path + "/" + str(dir_image[:-4]) + '_Affine_transfor_img' + str(index) + '.png', Affine_transfor_img)
res_rotate = random_rotate_img(img, 20, 70)
cv2.imwrite(child_path +"/" +str(dir_image[:-4])+ '_rotate_img' + str(index) + '.png', res_rotate[0])
HSV_IMG = random_hsv_transform(img,2,0.3,0.6)
cv2.imwrite(child_path +"/" +str(dir_image[:-4])+'_HSV_IMG' + str(index) + '.png', HSV_IMG)
GAMMA_IMG = random_gamma_transform(img,0.3)
cv2.imwrite(child_path +"/" +str(dir_image[:-4])+'_GAMMA_IMG' + str(index) + '.png', GAMMA_IMG)
res_flip = random_flip_img(img)
cv2.imwrite(child_path +"/" +str(dir_image[:-4])+'_flip_img'+ str(index) + '.png', res_flip)
# SP_NoiseImg = SaltAndPepper(img)
# cv2.imwrite(child_path + "/" + str(dir_image[:-4]) + '_SP_Noise_img' + str(index) + '.png', SP_NoiseImg)
G_Noiseimg = gaussian_noise(img)
cv2.imwrite(child_path + "/" + str(dir_image[:-4]) + '_G_Noise_img' + str(index) + '.png', G_Noiseimg)
index+=1
开启多线程
import os
import cv2
import random
import time
import numpy as np
import threading
from scipy import *
def Affine_transformation(img):
"""
:param img:
:return:
"""
# pointA = np.float32([[70, 30], [200, 50], [20,180]])
# pointB = np.float32([[30, 80], [200, 50], [80, 220]])
h, w = img.shape[:2]
points1 = np.float32([[50, 50], [200, 50], [50, 200]])
points2 = np.float32([[random.choice([30,40]), random.choice([70,80])],
[200, 50], [random.choice([70,80]), random.choice([210,220])]])
matrix = cv2.getAffineTransform(points1, points2)
Affine_transfor_img = cv2.warpAffine(img, matrix, (h, w))
return Affine_transfor_img
def random_rotate_img(img, min_angle, max_angle):
'''
random rotation an image
:param img: image to be rotated
:param min_angle: min angle to rotate
:param max_angle: max angle to rotate
:return: image after random rotated
'''
if not isinstance(img, list):
img = [img]
angle = random.randint(min_angle, max_angle)
center = (img[0].shape[0] / 2, img[0].shape[1] / 2)
rot_matrix = cv2.getRotationMatrix2D(center, angle, scale=1.0)
res = []
for img_inst in img:
img_inst = cv2.warpAffine(img_inst, rot_matrix, dsize=img_inst.shape[:2], borderMode=cv2.BORDER_CONSTANT)
res.append(img_inst)
if len(res) == 0:
res = res[0]
return res
def random_hsv_transform(img, hue_vari, sat_vari, val_vari):
"""
:param img:
:param hue_vari: 色调变化比例范围(0,360)
:param sat_vari: 饱和度变化比例范围(0,1)
:param val_vari: 明度变化比例范围(0,1)
:return:
"""
hue_delta = np.random.randint(-hue_vari, hue_vari)
sat_mult = 1 + np.random.uniform(-sat_vari, sat_vari)
val_mult = 1 + np.random.uniform(-val_vari, val_vari)
img_hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV).astype(np.float)
img_hsv[:, :, 0] = (img_hsv[:, :, 0] + hue_delta) % 180
img_hsv[:, :, 1] *= sat_mult
img_hsv[:, :, 2] *= val_mult
img_hsv[img_hsv > 255] = 255
return cv2.cvtColor(np.round(img_hsv).astype(np.uint8), cv2.COLOR_HSV2BGR)
def random_gamma_transform(img, gamma_vari):
"""
:param img:
:param gamma_vari:
:return:
"""
log_gamma_vari = np.log(gamma_vari)
alpha = np.random.uniform(-log_gamma_vari, log_gamma_vari)
gamma = np.exp(alpha)
gamma_table = [np.power(x / 255.0, gamma) * 255.0 for x in range(256)]
gamma_table = np.round(np.array(gamma_table)).astype(np.uint8)
return cv2.LUT(img, gamma_table)
def random_flip_img(img):
"""
0 = X axis, 1 = Y axis, -1 = both
:param img:
:return:
"""
flip_val = [0,1,-1]
random_flip_val = random.choice(flip_val)
res = cv2.flip(img, random_flip_val)
return res
def SaltAndPepper(img): #定义添加椒盐噪声的函数
"""
:param img:
:return:
"""
SP_NoiseImg=img
percetage = np.random.randint(3) * 0.1
SP_NoiseNum=int(percetage*img.shape[0]*img.shape[1])
for i in range(SP_NoiseNum):
randX=random.random_integers(0,img.shape[0]-1)
randY=random.random_integers(0,img.shape[1]-1)
if random.random_integers(0,1)==0:
SP_NoiseImg[randX,randY]=0
else:
SP_NoiseImg[randX,randY]=255
return SP_NoiseImg
def clamp(pv): #防止像素溢出
if pv > 255:
return 255
if pv < 0:
return 0
else:
return pv
def gaussian_noise(image): # 加高斯噪声
"""
:param image:
:return:
"""
h, w, c = image.shape
for row in range(h):
for col in range(w):
s = np.random.normal(0, 20, 3)
b = image[row, col, 0] # blue
g = image[row, col, 1] # green
r = image[row, col, 2] # red
image[row, col, 0] = clamp(b + s[0])
image[row, col, 1] = clamp(g + s[1])
image[row, col, 2] = clamp(r + s[2])
return image
def get_img(input_dir):
img_path_list = []
for (root_path,dirname,filenames) in os.walk(input_dir):
for filename in filenames:
img_path = root_path+"/"+filename
img_path_list.append(img_path)
return img_path_list
def start0(input_dir):
index = 1
img_path_list = get_img(input_dir)
for img_path in img_path_list:
Suffix_name = ['.png', '.jpg', '.jpeg']
if img_path.endswith(tuple(Suffix_name)):
print("线程1：处理图像",img_path)
img = cv2.imread(img_path)
GAMMA_IMG = random_gamma_transform(img, 0.3)
cv2.imwrite(img_path[:-4] + '_GAMMA_IMG' + str(index) + '.png', GAMMA_IMG)
res_flip = random_flip_img(img)
cv2.imwrite(img_path[:-4] + '_flip_img' + str(index) + '.png', res_flip)
G_Noiseimg = gaussian_noise(img)
cv2.imwrite(img_path[:-4] + '_G_Noise_img' + str(index) + '.png', G_Noiseimg)
index += 1
def start1(input_dir):
index = 1
img_path_list = get_img(input_dir)
for img_path in img_path_list:
Suffix_name = ['.png', '.jpg', '.jpeg']
if img_path.endswith(tuple(Suffix_name)):
print("线程2：处理图像", img_path)
img = cv2.imread(img_path)
Affine_transfor_img = Affine_transformation(img)
cv2.imwrite(img_path[:-4] + '_Affine_transfor_img' + str(index) + '.png',Affine_transfor_img)
res_rotate = random_rotate_img(img, 20, 70)
cv2.imwrite(img_path[:-4] + '_rotate_img' + str(index) + '.png', res_rotate[0])
HSV_IMG = random_hsv_transform(img, 2, 0.3, 0.6)
cv2.imwrite(img_path[:-4] + '_HSV_IMG' + str(index) + '.png', HSV_IMG)
index += 1
input_dir = "./All_faces"
threads = []
t1 = threading.Thread(target=start0,args=(input_dir,))
threads.append(t1)
t2 = threading.Thread(target=start1,args=(input_dir,))
threads.append(t2)
if __name__ == '__main__':
start_time = time.time()
for t in threads:
t.setDaemon(True)
t.start()
for t in threads:
t.join()
end_time = time.time()
print("done! \n Total Spend time：",str((end_time - start_time)/60)[0:6]+"分钟")
扫一扫
22万+
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
