252 lines
9.8 KiB
Python
252 lines
9.8 KiB
Python
import argparse
|
||
import math
|
||
import cv2
|
||
|
||
from typing import List
|
||
|
||
import numpy as np
|
||
from alibabacloud_imageseg20191230.client import Client as imageseg20191230Client
|
||
from alibabacloud_facebody20191230.client import Client as facebody20191230Client
|
||
from alibabacloud_tea_openapi import models as open_api_models
|
||
from alibabacloud_facebody20191230 import models as facebody_20191230_models
|
||
from alibabacloud_imageseg20191230 import models as imageseg_20191230_models
|
||
from alibabacloud_tea_util import models as util_models
|
||
from alibabacloud_tea_util.client import Client as UtilClient
|
||
from viapi.fileutils import FileUtils
|
||
|
||
import urllib.request
|
||
|
||
|
||
class Sample:
|
||
def __init__(self):
|
||
pass
|
||
|
||
@staticmethod
|
||
def create_client(
|
||
access_key_id: str,
|
||
access_key_secret: str,
|
||
) -> (imageseg20191230Client, facebody20191230Client):
|
||
"""
|
||
使用AK&SK初始化账号Client
|
||
@param access_key_id:
|
||
@param access_key_secret:
|
||
@return: Client
|
||
@throws Exception
|
||
"""
|
||
seg_config = open_api_models.Config(
|
||
# 您的 AccessKey ID,
|
||
access_key_id=access_key_id,
|
||
# 您的 AccessKey Secret,
|
||
access_key_secret=access_key_secret
|
||
)
|
||
face_config = open_api_models.Config(
|
||
# 您的 AccessKey ID,
|
||
access_key_id=access_key_id,
|
||
# 您的 AccessKey Secret,
|
||
access_key_secret=access_key_secret
|
||
)
|
||
# 访问的域名
|
||
seg_config.endpoint = f'imageseg.cn-shanghai.aliyuncs.com'
|
||
face_config.endpoint = f'facebody.cn-shanghai.aliyuncs.com'
|
||
return (imageseg20191230Client(seg_config), facebody20191230Client(face_config))
|
||
|
||
def id_pohto_demo(
|
||
args: List[str],
|
||
) -> None:
|
||
assert (args is not None and len(args.param) > 2), "parameters wrong, use -h for details!"
|
||
[sc_file, color, of_file] = [i for i in args.param]
|
||
|
||
# set id photo size
|
||
size = [295, 413]
|
||
|
||
# client init
|
||
seg_client, face_client = Sample.create_client(args.access_key_id, args.access_key_secret)
|
||
|
||
# generation source image url
|
||
input_url = generate_url(sc_file, args.access_key_id, args.access_key_secret)
|
||
|
||
# init segment_body_request
|
||
segment_body_request = imageseg_20191230_models.SegmentBodyRequest(
|
||
image_url=input_url
|
||
)
|
||
|
||
# init detect_face_request
|
||
detect_face_request = facebody_20191230_models.DetectFaceRequest(
|
||
image_url=input_url
|
||
)
|
||
runtime = util_models.RuntimeOptions()
|
||
try:
|
||
# 通过人脸SDK获取人脸105关键点信息
|
||
face_result = face_client.detect_face_with_options(detect_face_request, runtime)
|
||
landmarks = face_result.body.data.landmarks
|
||
landmarks = np.array(landmarks)
|
||
|
||
# 通过分割SDK获取分割png图
|
||
seg_result = seg_client.segment_body_with_options(segment_body_request, runtime)
|
||
print(seg_result.body.data.image_url)
|
||
rqt = urllib.request.urlopen(seg_result.body.data.image_url)
|
||
seg_img = np.asarray(bytearray(rqt.read()), dtype="uint8")
|
||
seg_img = cv2.imdecode(seg_img, cv2.IMREAD_UNCHANGED)
|
||
|
||
# 根据人脸关键点进行人脸对齐，返回旋转后图像、面部中心点、旋转角度、旋转后新的landmarks值
|
||
rotated_img, eye_center, angle, landmarks = align_face(seg_img, landmarks)
|
||
|
||
# 输入旋转后图片、预期的证件照分辨率、landmarks、背景色，返回半身像结果
|
||
png_img = corp_halfbody(rotated_img, landmarks, size)
|
||
|
||
# 证件照背景色替换
|
||
colors = {'red': (0, 0, 255, 255), 'blue': (255, 0, 0, 255), 'white': (255, 255, 255, 255)}
|
||
if type(color) is str:
|
||
color = colors[color]
|
||
rst_img = np.zeros((size[1], size[0], 3)) + color[0:3]
|
||
rst_img = image_merge_background(png_img[:, :, 0:3], png_img, rst_img)
|
||
|
||
# 保存结果到本地
|
||
cv2.imwrite(of_file, rst_img)
|
||
except Exception as error:
|
||
print(error.message)
|
||
# 如有需要，请打印 error
|
||
UtilClient.assert_as_string(error.message)
|
||
|
||
def align_face(image_array, landmarks):
|
||
""" align faces according to eyes position
|
||
:param image_array: numpy array of a single image
|
||
:param landmarks: dict of landmarks for facial parts as keys and tuple of coordinates as values
|
||
:return:
|
||
rotated_img: numpy array of aligned image
|
||
eye_center: tuple of coordinates for eye center
|
||
angle: degrees of rotation
|
||
"""
|
||
landmarks = np.resize(landmarks, (105, 2))
|
||
|
||
# get list landmarks of left and right eye
|
||
left_eye = landmarks[24:39]
|
||
right_eye = landmarks[40:55]
|
||
left_eye_center = np.mean(left_eye, axis=0).astype("int32")
|
||
right_eye_center = np.mean(right_eye, axis=0).astype("int32")
|
||
dy = right_eye_center[1] - left_eye_center[1]
|
||
dx = right_eye_center[0] - left_eye_center[0]
|
||
angle = math.atan2(dy, dx) * 180. / math.pi
|
||
# calculate the center of 2 eyes
|
||
eye_center = (int(left_eye_center[0] + right_eye_center[0]) // 2,
|
||
int(left_eye_center[1] + right_eye_center[1]) // 2)
|
||
# at the eye_center, rotate the image by the angle
|
||
rotate_matrix = cv2.getRotationMatrix2D(eye_center, angle, scale=1)
|
||
rotated_img = cv2.warpAffine(image_array, rotate_matrix, (image_array.shape[1], image_array.shape[0]))
|
||
|
||
rotated_landmarks = []
|
||
for landmark in landmarks:
|
||
rotated_landmark = rotate(origin=eye_center, point=landmark, angle=angle, row=image_array.shape[0])
|
||
rotated_landmarks.append(rotated_landmark)
|
||
return rotated_img, eye_center, angle, rotated_landmarks
|
||
|
||
def rotate(origin, point, angle, row):
|
||
""" rotate coordinates in image coordinate system
|
||
:param origin: tuple of coordinates,the rotation center
|
||
:param point: tuple of coordinates, points to rotate
|
||
:param angle: degrees of rotation
|
||
:param row: row size of the image
|
||
:return: rotated coordinates of point
|
||
"""
|
||
x1, y1 = point
|
||
x2, y2 = origin
|
||
y1 = row - y1
|
||
y2 = row - y2
|
||
angle = math.radians(angle)
|
||
x = x2 + math.cos(angle) * (x1 - x2) - math.sin(angle) * (y1 - y2)
|
||
y = y2 + math.sin(angle) * (x1 - x2) + math.cos(angle) * (y1 - y2)
|
||
y = row - y
|
||
return int(x), int(y)
|
||
|
||
def corp_halfbody(image_array, landmarks, size):
|
||
""" crop face according to eye,mouth and chin position
|
||
:param image_array: numpy array of a single image
|
||
:param size: single int value, size for w and h after crop
|
||
:param landmarks: dict of landmarks for facial parts as keys and tuple of coordinates as values
|
||
:return:
|
||
cropped_img: numpy array of cropped image
|
||
left, top: left and top coordinates of cropping
|
||
"""
|
||
|
||
crop_size = [0, 0, size[1], size[0]]
|
||
|
||
scal = size[1] / 4 / abs(landmarks[98][1] - landmarks[56][1])
|
||
image = cv2.resize(image_array, np.multiply((image_array.shape[0:2])[::-1], scal).astype(int))
|
||
landmarks = np.multiply(np.array(landmarks), scal)
|
||
|
||
x_center = landmarks[98][0]
|
||
crop_size[0:2] = [x_center - size[0] / 2, x_center + size[0] / 2]
|
||
|
||
y_center = (landmarks[98][1] + landmarks[56][1]) / 2
|
||
crop_size[2:4] = [y_center - size[1] / 2, y_center + size[1] / 2]
|
||
|
||
left, right, top, bottom = [round(i) for i in crop_size]
|
||
left = max(0, left)
|
||
top = max(0, top)
|
||
right = min(image.shape[1], right)
|
||
bottom = min(image.shape[0], bottom)
|
||
|
||
cropped_img = image[top:bottom, left:right]
|
||
|
||
left, right, top, bottom = [round(i) for i in crop_size]
|
||
bottom = size[1]
|
||
top = size[1] - cropped_img.shape[0]
|
||
left = -min(0, left)
|
||
right = min(cropped_img.shape[1] + left, size[0])
|
||
|
||
|
||
png_img = np.zeros((size[1], size[0], 4))
|
||
png_img[top:bottom, left:right] = cropped_img[:, :]
|
||
return png_img
|
||
|
||
def image_merge_background(sc_image, png_image, bg_image):
|
||
""" merge foreground to background image
|
||
:param sc_image: numpy array of the source image
|
||
:param png_image: numpy array of the segmented result with same size of sc_image
|
||
:param bg_image: numpy array of the background image with same size of sc_image
|
||
:return:
|
||
rst_image: numpy array of merged image
|
||
"""
|
||
assert (sc_image is not None and png_image is not None and bg_image is not None), "read image input error!"
|
||
h, w, c = sc_image.shape
|
||
|
||
# keep sc_image, png_image and bg_image same size
|
||
viapi_image = cv2.resize(png_image, (w, h))
|
||
bg_image = cv2.resize(bg_image, (w, h))
|
||
if len(viapi_image.shape) == 2:
|
||
mask = viapi_image[:, :, np.newaxis]
|
||
elif viapi_image.shape[2] == 4:
|
||
mask = viapi_image[:, :, 3:4]
|
||
elif viapi_image.shape[2] == 3:
|
||
mask = viapi_image[:, :, 0:1]
|
||
else:
|
||
raise Exception("invalid image mask!")
|
||
mask = mask / 255.0
|
||
|
||
# merge background
|
||
sc_image = sc_image.astype(float)
|
||
bg_image = bg_image.astype(float)
|
||
rst_image = (sc_image - bg_image) * mask + bg_image
|
||
rst_image = np.clip(rst_image, 0, 255)
|
||
return rst_image
|
||
|
||
# see https://help.aliyun.com/document_detail/155645.html for more detail
|
||
def generate_url(image_path, access_key_id, access_key_secret):
|
||
file_utils = FileUtils(access_key_id, access_key_secret)
|
||
oss_url = file_utils.get_oss_url(image_path, "jpeg", True)
|
||
return oss_url
|
||
|
||
def define_options():
|
||
parser = argparse.ArgumentParser(description='Human segmentation examples.')
|
||
parser.add_argument('-i', '--access_key_id', default="LTAI5tE6gkXQqJBbCxFfo4hv", help="oss access key id")
|
||
parser.add_argument('-s', '--access_key_secret', default="pziJRQfXlev3vEIY4dHuB6C7E73Uv6", help="access key secret")
|
||
parser.add_argument('-p', '--param', nargs=3
|
||
, default=['./data/source.jpeg', 'blue', 'result2.jpg']
|
||
, type=str
|
||
, help=('source_image_path background_color[red blue white] result_path'))
|
||
args = parser.parse_args()
|
||
return args
|
||
|
||
if __name__ == '__main__':
|
||
args = define_options()
|
||
id_pohto_demo(args)
|