文章目录
1、功能描述
利用液化效果实现瘦脸美颜
交互式的液化效果原理来自 Gustafsson A. Interactive image warping[D]. , 1993.
2、原理分析
上面描述很清晰了，鼠标初始在 C，也即形变范围的圆心在 C，形变半径 r m a x r_{max} rmax，形变方向 C→M，
圆圈内原始 U 位置会被形变到 X，可以简单直白理解为拉伸后 U 位置的值给了 X 位置，此时 U 位置空置了，需要插值
插值公示 93 年的论文中直接给出了，我们尝试 coding
这里还涉及到插值，我们回顾下比较常见的双线性插值原理
3、代码实现
导入必要的库函数
python
import dlib
import cv2
import numpy as np
import math载入人脸检测器和人脸关键点检测模型
python
predictor_path = "./shape_predictor_68_face_landmarks.dat"
# 使用dlib自带的frontal_face_detector作为我们的特征提取器
detector = dlib.get_frontal_face_detector() # 人脸检测器
predictor = dlib.shape_predictor(predictor_path) # 关键点检测模型读入图片，调用 face_thin_auto 函数，实现瘦脸
python
def main():
src = cv2.imread(r'./1.jpg') # (1546, 1236, 3)
# cv2.imshow('src', src)
face_thin_auto(src)
cv2.waitKey(0)
if __name__ == '__main__':
main()
看看 face_thin_auto 函数的实现细节
python
def face_thin_auto(src):
landmarks = landmark_dec_dlib_fun(src)
point_img = src.copy()
for index, landmark in enumerate(landmarks[0]):
cv2.circle(point_img, center=np.array(landmark)[0], radius=5, color=(255, 0, 0), thickness=-1)
cv2.putText(point_img, str(index), org=(landmark[0,0]-30, landmark[0,1]),
fontFace=cv2.FONT_HERSHEY_TRIPLEX,fontScale=0.5, color=(0,255,0))
cv2.imwrite("point.jpg", point_img)
# 如果未检测到人脸关键点，就不进行瘦脸
if len(landmarks) == 0:
print("not detect face keypoint")
return
thin_image = src
landmarks_node = landmarks[0]
endPt = landmarks_node[16] # matrix([[753, 450]])
for index in range(3, 14, 2):
start_landmark = landmarks_node[index]
end_landmark = landmarks_node[index + 2]
r = math.sqrt((start_landmark[0, 0] - end_landmark[0, 0]) **2 +
(start_landmark[0, 1] - end_landmark[0, 1]) **2)
thin_image = localTranslationWarp(thin_image, start_landmark[0, 0],
start_landmark[0, 1], endPt[0, 0], endPt[0, 1], r)
# 显示
# cv2.imshow('thin', thin_image)
cv2.imwrite(r'./thin.jpg', thin_image)landmark_dec_dlib_fun 检测人脸和人脸关键点，实现如下
python
def landmark_dec_dlib_fun(img_src):
img_gray = cv2.cvtColor(img_src, cv2.COLOR_BGR2GRAY)
cv2.imwrite("gray.jpg", img_gray)
land_marks = []
rects = detector(img_gray, 0) # 人脸检测，[[(336, 286) (782, 732)]]
plot_img = img_src.copy()
for i in range(len(rects)): # 遍历检测到的人脸
cv2.rectangle(plot_img, (rects[i].left(), rects[i].top()), (rects[i].right(), rects[i].bottom()),
color=(0,255,0), thickness=10)
land_marks_node = np.matrix([[p.x, p.y] for p in predictor(img_gray, rects[i]).parts()])
land_marks.append(land_marks_node)
cv2.imwrite("face_det.jpg", plot_img)
return land_marks先把图片变成灰度图，然后人脸检测，绘制人脸检测结果，人脸关键点检测，返回关键点坐标
face_thin_auto 函数接下来绘制人脸关键点，一共 68 个
遍历关键点，3，5，7，9，11，13
也即 C = 3，5，7，9，11，13，M = 16
r m a x r_{max} rmax 为关键点 3-5 的距离，5-7 的距离，7-9 的距离，9-11 的距离，11-13 的距离，13-15 的距离
调用 localTranslationWarp 求瘦脸后的图片，
python
def localTranslationWarp(srcImg, startX, startY, endX, endY, radius):
ddradius = float(radius * radius)
copyImg = srcImg.copy()
# 计算公式中的|m-c|^2
ddmc = (endX - startX) ** 2 + (endY - startY) ** 2
H, W, C = srcImg.shape
for i in range(W):
for j in range(H):
# 计算该点是否在形变圆的范围之内
# 优化，第一步，直接判断是会在（startX,startY)的矩阵框中
if math.fabs(i - startX) > radius and math.fabs(j - startY) > radius:
continue # 不在 continue
distance = (i - startX) ** 2 + (j - startY) ** 2
if (distance < ddradius):
# 计算出（i,j）坐标的原坐标
# 计算公式中右边平方号里的部分
ratio = (ddradius - distance) / (ddradius - distance + ddmc)
ratio = ratio ** 2
# 映射原位置
UX = i - ratio * (endX - startX)
UY = j - ratio * (endY - startY)
# 根据双线性插值法得到UX，UY的值
value = BilinearInsert(srcImg, UX, UY)
# 改变当前 i ，j的值
copyImg[j, i] = value
return copyImglocalTranslationWarp 仅作用与以 C 为圆心， r m a x r_{max} rmax 范围内的像素点，像素点的坐标求法代入公式计算，值用插值求出
双线性插值实现
python
def BilinearInsert(src, ux, uy):
w, h, c = src.shape
if c == 3:
x1 = int(ux)
x2 = x1 + 1
y1 = int(uy)
y2 = y1 + 1
part1 = src[y1, x1].astype(float) * (float(x2) - ux) * (float(y2) - uy)
part2 = src[y1, x2].astype(float) * (ux - float(x1)) * (float(y2) - uy)
part3 = src[y2, x1].astype(float) * (float(x2) - ux) * (uy - float(y1))
part4 = src[y2, x2].astype(float) * (ux - float(x1)) * (uy - float(y1))
insertValue = part1 + part2 + part3 + part4
return insertValue.astype(np.int8)我们看看
python
part1 = src[y1, x1].astype(float) * (float(x2) - ux) * (float(y2) - uy)
part2 = src[y1, x2].astype(float) * (ux - float(x1)) * (float(y2) - uy)
part3 = src[y2, x1].astype(float) * (float(x2) - ux) * (uy - float(y1))
part4 = src[y2, x2].astype(float) * (ux - float(x1)) * (uy - float(y1))对应
f ( Q 11 ) ∗ x 2 − x x 2 − x 1 ∗ y 2 − y y 2 − y 1 = f ( Q 11 ) ∗ ( x 2 − x ) ∗ ( y 2 − y ) f(Q_{11}) * \frac{x_2 - x}{x_2- x_1} * \frac{y_2 - y}{y_2- y_1} = f(Q_{11}) * (x_2 - x) * (y_2 - y) f(Q11)∗x2−x1x2−x∗y2−y1y2−y=f(Q11)∗(x2−x)∗(y2−y)
f ( Q 21 ) ∗ x − x 1 x 2 − x 1 ∗ y 2 − y y 2 − y 1 = f ( Q 21 ) ∗ ( x − x 1 ) ∗ ( y 2 − y ) f(Q_{21}) * \frac{x - x_1}{x_2- x_1} * \frac{y_2 - y}{y_2- y_1} = f(Q_{21}) * (x - x_1) * (y_2 - y) f(Q21)∗x2−x1x−x1∗y2−y1y2−y=f(Q21)∗(x−x1)∗(y2−y)
f ( Q 12 ) ∗ x 2 − x x 2 − x 1 ∗ y − y 1 y 2 − y 1 = f ( Q 12 ) ∗ ( x 2 − x ) ∗ ( y − y 1 ) f(Q_{12}) * \frac{x_2 - x}{x_2- x_1} * \frac{y - y_1}{y_2- y_1} = f(Q_{12}) * (x_2 - x) * (y - y_1) f(Q12)∗x2−x1x2−x∗y2−y1y−y1=f(Q12)∗(x2−x)∗(y−y1)
f ( Q 22 ) ∗ x − x 1 x 2 − x 1 ∗ y − y 1 y 2 − y 1 = f ( Q 22 ) ∗ ( x − x 1 ) ∗ ( y − y 1 ) f(Q_{22}) * \frac{x - x_1}{x_2- x_1} * \frac{y - y_1}{y_2- y_1} = f(Q_{22}) * (x - x_1) * (y - y_1) f(Q22)∗x2−x1x−x1∗y2−y1y−y1=f(Q22)∗(x−x1)∗(y−y1)
4、效果展示
输入
输出
再明显一点试试
输入
输出
肉眼看不太明显，对比工具看比较明显
缩小下图片的输入分辨率
效果会明显一些
5、完整代码
python
import dlib
import cv2
import numpy as np
import math
predictor_path = "./shape_predictor_68_face_landmarks.dat"
# 使用dlib自带的frontal_face_detector作为我们的特征提取器
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(predictor_path)
def landmark_dec_dlib_fun(img_src):
img_gray = cv2.cvtColor(img_src, cv2.COLOR_BGR2GRAY)
cv2.imwrite("gray.jpg", img_gray)
land_marks = []
rects = detector(img_gray, 0) # 人脸检测，[[(336, 286) (782, 732)]]
plot_img = img_src.copy()
for i in range(len(rects)): # 遍历检测到的人脸
cv2.rectangle(plot_img, (rects[i].left(), rects[i].top()), (rects[i].right(), rects[i].bottom()),
color=(0,255,0), thickness=10)
land_marks_node = np.matrix([[p.x, p.y] for p in predictor(img_gray, rects[i]).parts()])
land_marks.append(land_marks_node)
cv2.imwrite("face_det.jpg", plot_img)
return land_marks
'''
方法： Interactive Image Warping 局部平移算法
'''
def localTranslationWarp(srcImg, startX, startY, endX, endY, radius):
ddradius = float(radius * radius)
copyImg = srcImg.copy()
# 计算公式中的|m-c|^2
ddmc = (endX - startX) ** 2 + (endY - startY) ** 2
H, W, C = srcImg.shape
for i in range(W):
for j in range(H):
# 计算该点是否在形变圆的范围之内
# 优化，第一步，直接判断是会在（startX,startY)的矩阵框中
if math.fabs(i - startX) > radius and math.fabs(j - startY) > radius:
continue # 不在 continue
distance = (i - startX) ** 2 + (j - startY) ** 2
if (distance < ddradius):
# 计算出（i,j）坐标的原坐标
# 计算公式中右边平方号里的部分
ratio = (ddradius - distance) / (ddradius - distance + ddmc)
ratio = ratio ** 2
# 映射原位置
UX = i - ratio * (endX - startX)
UY = j - ratio * (endY - startY)
# 根据双线性插值法得到UX，UY的值
value = BilinearInsert(srcImg, UX, UY)
# 改变当前 i ，j的值
copyImg[j, i] = value
return copyImg
# 双线性插值法
def BilinearInsert(src, ux, uy):
w, h, c = src.shape
if c == 3:
x1 = int(ux)
x2 = x1 + 1
y1 = int(uy)
y2 = y1 + 1
part1 = src[y1, x1].astype(float) * (float(x2) - ux) * (float(y2) - uy)
part2 = src[y1, x2].astype(float) * (ux - float(x1)) * (float(y2) - uy)
part3 = src[y2, x1].astype(float) * (float(x2) - ux) * (uy - float(y1))
part4 = src[y2, x2].astype(float) * (ux - float(x1)) * (uy - float(y1))
insertValue = part1 + part2 + part3 + part4
return insertValue.astype(np.int8)
def face_thin_auto(src):
landmarks = landmark_dec_dlib_fun(src)
point_img = src.copy()
for index, landmark in enumerate(landmarks[0]):
cv2.circle(point_img, center=np.array(landmark)[0], radius=5, color=(255, 0, 0), thickness=-1)
cv2.putText(point_img, str(index), org=(landmark[0,0]-30, landmark[0,1]),
fontFace=cv2.FONT_HERSHEY_TRIPLEX,fontScale=0.5, color=(0,255,0))
cv2.imwrite("point.jpg", point_img)
# 如果未检测到人脸关键点，就不进行瘦脸
if len(landmarks) == 0:
print("not detect face keypoint")
return
thin_image = src
landmarks_node = landmarks[0]
endPt = landmarks_node[16] # matrix([[753, 450]])
for index in range(3, 14, 2):
start_landmark = landmarks_node[index]
end_landmark = landmarks_node[index + 2]
r = math.sqrt((start_landmark[0, 0] - end_landmark[0, 0]) **2 +
(start_landmark[0, 1] - end_landmark[0, 1]) **2)
thin_image = localTranslationWarp(thin_image, start_landmark[0, 0],
start_landmark[0, 1], endPt[0, 0], endPt[0, 1], r)
# 显示
# cv2.imshow('thin', thin_image)
cv2.imwrite(r'./thin.jpg', thin_image)
def main():
src = cv2.imread(r'./1.jpg') # (1546, 1236, 3)
# cv2.imshow('src', src)
face_thin_auto(src)
cv2.waitKey(0)
if __name__ == '__main__':
main()6、参考
-
shape_predictor_68_face_landmarks.dat.bz2
-
链接: https://pan.baidu.com/s/1gO_wqRAtWndGkUhZOSBw2Q?pwd=4enn
提取码: 4enn
-
-
