每到大型節假日,我們常會發現社交平臺都會提供生成頭像裝飾的小工具,很是新奇好玩。如果從技術的維度看,這類平臺/工具一般都是通過下面兩個方法給我們生成頭像裝飾的:
- 一是直接加裝飾,例如在頭像外面加邊框,在下面加logo等;
- 二是通過機器學習算法增加裝飾,例如增加一個聖誕帽等;
使用Serverless直接增加頭像裝飾
增加頭像裝飾的功能其實很容易實現,首先選擇一張圖片,上傳自己的頭像,然後函數部分進行圖像的合成,這一部分並沒有涉及到機器學習算法,僅僅是圖像合成相關算法。
通過用戶上傳的圖片,在指定位置增加預定圖片/用戶選擇的圖片作爲裝飾物進行添加:
- 將預定圖片/用戶選擇的圖片進行美化,此處僅是將其變成圓形:
def do_circle(base_pic):
icon_pic = Image.open(base_pic).convert("RGBA")
icon_pic = icon_pic.resize((500, 500), Image.ANTIALIAS)
icon_pic_x, icon_pic_y = icon_pic.size
temp_icon_pic = Image.new('RGBA', (icon_pic_x + 600, icon_pic_y + 600), (255, 255, 255))
temp_icon_pic.paste(icon_pic, (300, 300), icon_pic)
ima = temp_icon_pic.resize((200, 200), Image.ANTIALIAS)
size = ima.size
# 因爲是要圓形,所以需要正方形的圖片
r2 = min(size[0], size[1])
if size[0] != size[1]:
ima = ima.resize((r2, r2), Image.ANTIALIAS)
# 最後生成圓的半徑
r3 = 60
imb = Image.new('RGBA', (r3 * 2, r3 * 2), (255, 255, 255, 0))
pima = ima.load() # 像素的訪問對象
pimb = imb.load()
r = float(r2 / 2) # 圓心橫座標
for i in range(r2):
for j in range(r2):
lx = abs(i - r) # 到圓心距離的橫座標
ly = abs(j - r) # 到圓心距離的縱座標
l = (pow(lx, 2) + pow(ly, 2)) ** 0.5 # 三角函數 半徑
if l < r3:
pimb[i - (r - r3), j - (r - r3)] = pima[i, j]
return imb
- 添加該裝飾到用戶頭像上:
def add_decorate(base_pic):
try:
base_pic = "./base/%s.png" % (str(base_pic))
user_pic = Image.open("/tmp/picture.png").convert("RGBA")
temp_basee_user_pic = Image.new('RGBA', (440, 440), (255, 255, 255))
user_pic = user_pic.resize((400, 400), Image.ANTIALIAS)
temp_basee_user_pic.paste(user_pic, (20, 20))
temp_basee_user_pic.paste(do_circle(base_pic), (295, 295), do_circle(base_pic))
temp_basee_user_pic.save("/tmp/output.png")
return True
except Exception as e:
print(e)
return False
- 除此之外,爲了方便本地測試,項目增加了
test()
方法模擬API網關傳遞的數據:
def test():
with open("test.png", 'rb') as f:
image = f.read()
image_base64 = str(base64.b64encode(image), encoding='utf-8')
event = {
"requestContext": {
"serviceId": "service-f94sy04v",
"path": "/test/{path}",
"httpMethod": "POST",
"requestId": "c6af9ac6-7b61-11e6-9a41-93e8deadbeef",
"identity": {
"secretId": "abdcdxxxxxxxsdfs"
},
"sourceIp": "14.17.22.34",
"stage": "release"
},
"headers": {
"Accept-Language": "en-US,en,cn",
"Accept": "text/html,application/xml,application/json",
"Host": "service-3ei3tii4-251000691.ap-guangzhou.apigateway.myqloud.com",
"User-Agent": "User Agent String"
},
"body": "{\"pic\":\"%s\", \"base\":\"1\"}" % image_base64,
"pathParameters": {
"path": "value"
},
"queryStringParameters": {
"foo": "bar"
},
"headerParameters": {
"Refer": "10.0.2.14"
},
"stageVariables": {
"stage": "release"
},
"path": "/test/value",
"queryString": {
"foo": "bar",
"bob": "alice"
},
"httpMethod": "POST"
}
print(main_handler(event, None))
if __name__ == "__main__":
test()
- 爲了讓函數有同一個返回規範,此處增加統一返回的函數:
def return_msg(error, msg):
return_data = {
"uuid": str(uuid.uuid1()),
"error": error,
"message": msg
}
print(return_data)
return return_data
- 最後是塗口函數的寫法:
import base64, json
from PIL import Image
import uuid
def main_handler(event, context):
try:
print("將接收到的base64圖像轉爲pic")
imgData = base64.b64decode(json.loads(event["body"])["pic"].split("base64,")[1])
with open('/tmp/picture.png', 'wb') as f:
f.write(imgData)
basePic = json.loads(event["body"])["base"]
addResult = add_decorate(basePic)
if addResult:
with open("/tmp/output.png", "rb") as f:
base64Data = str(base64.b64encode(f.read()), encoding='utf-8')
return return_msg(False, {"picture": base64Data})
else:
return return_msg(True, "飾品添加失敗")
except Exception as e:
return return_msg(True, "數據處理異常: %s" % str(e))
完成後端圖像合成功能,製作前端頁面:
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>2020頭像大變樣 - 頭像SHOW - 自豪的採用騰訊雲Serverless架構!</title>
<meta name="viewport" content="width=device-width, initial-scale=1,maximum-scale=1,user-scalable=no">
<meta name="apple-mobile-web-app-capable" content="yes">
<meta name="apple-mobile-web-app-status-bar-style" content="black">
<script type="text/javascript">
thisPic = null
function getFileUrl(sourceId) {
var url;
thisPic = document.getElementById(sourceId).files.item(0)
if (navigator.userAgent.indexOf("MSIE") >= 1) { // IE
url = document.getElementById(sourceId).value;
} else if (navigator.userAgent.indexOf("Firefox") > 0) { // Firefox
url = window.URL.createObjectURL(document.getElementById(sourceId).files.item(0));
} else if (navigator.userAgent.indexOf("Chrome") > 0) { // Chrome
url = window.URL.createObjectURL(document.getElementById(sourceId).files.item(0));
}
return url;
}
function preImg(sourceId, targetId) {
var url = getFileUrl(sourceId);
var imgPre = document.getElementById(targetId);
imgPre.aaaaaa = url;
imgPre.style = "display: block;";
}
function clickChose() {
document.getElementById("imgOne").click()
}
function getNewPhoto() {
document.getElementById("result").innerText = "系統處理中,請稍後..."
var oFReader = new FileReader();
oFReader.readAsDataURL(thisPic);
oFReader.onload = function (oFREvent) {
var xmlhttp;
if (window.XMLHttpRequest) {
// IE7+, Firefox, Chrome, Opera, Safari 瀏覽器執行代碼
xmlhttp = new XMLHttpRequest();
} else {
// IE6, IE5 瀏覽器執行代碼
xmlhttp = new ActiveXObject("Microsoft.XMLHTTP");
}
xmlhttp.onreadystatechange = function () {
if (xmlhttp.readyState == 4 && xmlhttp.status == 200) {
if (JSON.parse(xmlhttp.responseText)["error"]) {
document.getElementById("result").innerText = JSON.parse(xmlhttp.responseText)["message"];
} else {
document.getElementById("result").innerText = "長按保存圖像";
document.getElementById("new_photo").aaaaaa = "data:image/png;base64," + JSON.parse(xmlhttp.responseText)["message"]["picture"];
document.getElementById("new_photo").style = "display: block;";
}
}
}
var url = " http://service-8d3fi753-1256773370.bj.apigw.tencentcs.com/release/new_year_add_photo_decorate"
var obj = document.getElementsByName("base");
var baseNum = "1"
for (var i = 0; i < obj.length; i++) {
console.log(obj[i].checked)
if (obj[i].checked) {
baseNum = obj[i].value;
}
}
xmlhttp.open("POST", url, true);
xmlhttp.setRequestHeader("Content-type", "application/json");
var postData = {
pic: oFREvent.target.result,
base: baseNum
}
xmlhttp.send(JSON.stringify(postData));
}
}
</script>
<!--標準mui.css-->
<link rel="stylesheet" href="./css/mui.min.css">
</head>
<body>
<h3 style="text-align: center; margin-top: 30px">2020頭像SHOW</h3>
<div class="mui-card">
<div class="mui-card-content">
<div class="mui-card-content-inner">
第一步:選擇一個你喜歡的圖片
</div>
</div>
<div class="mui-content">
<ul class="mui-table-view mui-grid-view mui-grid-9">
<li class="mui-table-view-cell mui-media mui-col-xs-4 mui-col-sm-3"><label>
<img aaaaaa="./base/1.png" width="100%"><input type="radio" name="base" value="1" checked></label></li>
<li class="mui-table-view-cell mui-media mui-col-xs-4 mui-col-sm-3"><label>
<img aaaaaa="./base/2.png" width="100%"><input type="radio" name="base" value="2"></label></li>
<li class="mui-table-view-cell mui-media mui-col-xs-4 mui-col-sm-3"><label>
<img aaaaaa="./base/11.png" width="100%"><input type="radio" name="base" value="11"></label></li>
<li class="mui-table-view-cell mui-media mui-col-xs-4 mui-col-sm-3"><label>
<img aaaaaa="./base/4.png" width="100%"><input type="radio" name="base" value="4"></label></li>
<li class="mui-table-view-cell mui-media mui-col-xs-4 mui-col-sm-3"><label>
<img aaaaaa="./base/5.png" width="100%"><input type="radio" name="base" value="5"></label></li>
<li class="mui-table-view-cell mui-media mui-col-xs-4 mui-col-sm-3"><label>
<img aaaaaa="./base/6.png" width="100%"><input type="radio" name="base" value="6"></label></li>
<li class="mui-table-view-cell mui-media mui-col-xs-4 mui-col-sm-3"><label>
<img aaaaaa="./base/12.png" width="100%"><input type="radio" name="base" value="12"></label></li>
<li class="mui-table-view-cell mui-media mui-col-xs-4 mui-col-sm-3"><label>
<img aaaaaa="./base/8.png" width="100%"><input type="radio" name="base" value="8"></label></li>
<li class="mui-table-view-cell mui-media mui-col-xs-4 mui-col-sm-3"><label>
<img aaaaaa="./base/3.png" width="100%"><input type="radio" name="base" value="3"></label></li>
</ul>
</div>
</div>
<div class="mui-card">
<div class="mui-card-content">
<div class="mui-card-content-inner">
第二步:上傳一張你的頭像
</div>
<div>
<form>
<input type="file" name="imgOne" id="imgOne" onchange="preImg(this.id, 'photo')" style="display: none;"
accept="image/*">
<center style="margin-bottom: 10px">
<input type="button" value="點擊此處上傳頭像" onclick="clickChose()"/>
<img id="photo" aaaaaa="" width="300px" , height="300px" style="display: none;"/>
</center>
</form>
</div>
</div>
</div>
<div class="mui-card">
<div class="mui-card-content">
<div class="mui-card-content-inner">
第三步:點擊生成按鈕獲取新年頭像
</div>
<div>
<center style="margin-bottom: 10px">
<input type="button" value="生成新年頭像" onclick="getNewPhoto()"/>
<p id="result"></p>
<img id="new_photo" aaaaaa="" width="300px" , height="300px" style="display: none;"/>
</center>
</div>
</div>
</div>
<p style="text-align: center">
本項目自豪的<br>通過Serverless Framework<br>搭建在騰訊雲SCF上
</p>
</body>
</html>
完成之後:
new_year_add_photo_decorate:
component: "@serverless/tencent-scf"
inputs:
name: myapi_new_year_add_photo_decorate
codeUri: ./new_year_add_photo_decorate
handler: index.main_handler
runtime: Python3.6
region: ap-beijing
description: 新年爲頭像增加飾品
memorySize: 128
timeout: 5
events:
- apigw:
name: serverless
parameters:
serviceId: service-8d3fi753
environment: release
endpoints:
- path: /new_year_add_photo_decorate
description: 新年爲頭像增加飾品
method: POST
enableCORS: true
param:
- name: pic
position: BODY
required: 'FALSE'
type: string
desc: 原始圖片
- name: base
position: BODY
required: 'FALSE'
type: string
desc: 飾品ID
myWebsite:
component: '@serverless/tencent-website'
inputs:
code:
src: ./new_year_add_photo_decorate/web
index: index.html
error: index.html
region: ap-beijing
bucketName: new-year-add-photo-decorate
完成之後就可以實現頭像加裝飾的功能,效果如下:
Serverless與人工智能聯手增加頭像裝飾
直接加裝飾的方式其實是可以在前端實現的,但是既然用到了後端服務和雲函數,那麼我們不妨就將人工智能與Serverless架構結果來實現一個增加裝飾的小工具。
實現這一功能的主要做法就是通過人工智能算法(此處是通過Dlib實現)進行人臉檢測:
print("dlib人臉關鍵點檢測器,正臉檢測")
predictorPath = "shape_predictor_5_face_landmarks.dat"
predictor = dlib.shape_predictor(predictorPath)
detector = dlib.get_frontal_face_detector()
dets = detector(img, 1)
此處的做法是隻檢測一張臉,檢測到即進行返回:
for d in dets:
x, y, w, h = d.left(), d.top(), d.right() - d.left(), d.bottom() - d.top()
print("關鍵點檢測,5個關鍵點")
shape = predictor(img, d)
print("選取左右眼眼角的點")
point1 = shape.part(0)
point2 = shape.part(2)
print("求兩點中心")
eyes_center = ((point1.x + point2.x) // 2, (point1.y + point2.y) // 2)
print("根據人臉大小調整帽子大小")
factor = 1.5
resizedHatH = int(round(rgbHat.shape[0] * w / rgbHat.shape[1] * factor))
resizedHatW = int(round(rgbHat.shape[1] * w / rgbHat.shape[1] * factor))
if resizedHatH > y:
resizedHatH = y - 1
print("根據人臉大小調整帽子大小")
resizedHat = cv2.resize(rgbHat, (resizedHatW, resizedHatH))
print("用alpha通道作爲mask")
mask = cv2.resize(a, (resizedHatW, resizedHatH))
maskInv = cv2.bitwise_not(mask)
print("帽子相對與人臉框上線的偏移量")
dh = 0
bgRoi = img[y + dh - resizedHatH:y + dh,
(eyes_center[0] - resizedHatW // 3):(eyes_center[0] + resizedHatW // 3 * 2)]
print("原圖ROI中提取放帽子的區域")
bgRoi = bgRoi.astype(float)
maskInv = cv2.merge((maskInv, maskInv, maskInv))
alpha = maskInv.astype(float) / 255
print("相乘之前保證兩者大小一致(可能會由於四捨五入原因不一致)")
alpha = cv2.resize(alpha, (bgRoi.shape[1], bgRoi.shape[0]))
bg = cv2.multiply(alpha, bgRoi)
bg = bg.astype('uint8')
print("提取帽子區域")
hat = cv2.bitwise_and(resizedHat, cv2.bitwise_not(maskInv))
print("相加之前保證兩者大小一致(可能會由於四捨五入原因不一致)")
hat = cv2.resize(hat, (bgRoi.shape[1], bgRoi.shape[0]))
print("兩個ROI區域相加")
addHat = cv2.add(bg, hat)
print("把添加好帽子的區域放回原圖")
img[y + dh - resizedHatH:y + dh,
(eyes_center[0] - resizedHatW // 3):(eyes_center[0] + resizedHatW // 3 * 2)] = addHat
return img
在Serverless架構下的完整代碼:
import cv2
import dlib
import base64
import json
def addHat(img, hat_img):
print("分離rgba通道,合成rgb三通道帽子圖,a通道後面做mask用")
r, g, b, a = cv2.split(hat_img)
rgbHat = cv2.merge((r, g, b))
print("dlib人臉關鍵點檢測器,正臉檢測")
predictorPath = "shape_predictor_5_face_landmarks.dat"
predictor = dlib.shape_predictor(predictorPath)
detector = dlib.get_frontal_face_detector()
dets = detector(img, 1)
print("如果檢測到人臉")
if len(dets) > 0:
for d in dets:
x, y, w, h = d.left(), d.top(), d.right() - d.left(), d.bottom() - d.top()
print("關鍵點檢測,5個關鍵點")
shape = predictor(img, d)
print("選取左右眼眼角的點")
point1 = shape.part(0)
point2 = shape.part(2)
print("求兩點中心")
eyes_center = ((point1.x + point2.x) // 2, (point1.y + point2.y) // 2)
print("根據人臉大小調整帽子大小")
factor = 1.5
resizedHatH = int(round(rgbHat.shape[0] * w / rgbHat.shape[1] * factor))
resizedHatW = int(round(rgbHat.shape[1] * w / rgbHat.shape[1] * factor))
if resizedHatH > y:
resizedHatH = y - 1
print("根據人臉大小調整帽子大小")
resizedHat = cv2.resize(rgbHat, (resizedHatW, resizedHatH))
print("用alpha通道作爲mask")
mask = cv2.resize(a, (resizedHatW, resizedHatH))
maskInv = cv2.bitwise_not(mask)
print("帽子相對與人臉框上線的偏移量")
dh = 0
bgRoi = img[y + dh - resizedHatH:y + dh,
(eyes_center[0] - resizedHatW // 3):(eyes_center[0] + resizedHatW // 3 * 2)]
print("原圖ROI中提取放帽子的區域")
bgRoi = bgRoi.astype(float)
maskInv = cv2.merge((maskInv, maskInv, maskInv))
alpha = maskInv.astype(float) / 255
print("相乘之前保證兩者大小一致(可能會由於四捨五入原因不一致)")
alpha = cv2.resize(alpha, (bgRoi.shape[1], bgRoi.shape[0]))
bg = cv2.multiply(alpha, bgRoi)
bg = bg.astype('uint8')
print("提取帽子區域")
hat = cv2.bitwise_and(resizedHat, cv2.bitwise_not(maskInv))
print("相加之前保證兩者大小一致(可能會由於四捨五入原因不一致)")
hat = cv2.resize(hat, (bgRoi.shape[1], bgRoi.shape[0]))
print("兩個ROI區域相加")
addHat = cv2.add(bg, hat)
print("把添加好帽子的區域放回原圖")
img[y + dh - resizedHatH:y + dh,
(eyes_center[0] - resizedHatW // 3):(eyes_center[0] + resizedHatW // 3 * 2)] = addHat
return img
def main_handler(event, context):
try:
print("將接收到的base64圖像轉爲pic")
imgData = base64.b64decode(json.loads(event["body"])["pic"])
with open('/tmp/picture.png', 'wb') as f:
f.write(imgData)
print("讀取帽子素材以及用戶頭像")
hatImg = cv2.imread("hat.png", -1)
userImg = cv2.imread("/tmp/picture.png")
output = addHat(userImg, hatImg)
cv2.imwrite("/tmp/output.jpg", output)
print("讀取頭像進行返回給用戶,以Base64返回")
with open("/tmp/output.jpg", "rb") as f:
base64Data = str(base64.b64encode(f.read()), encoding='utf-8')
return {
"picture": base64Data
}
except Exception as e:
return {
"error": str(e)
}
這樣,我們就完成了通過用戶上傳人物頭像進行增加聖誕帽的功能。
總結
傳統情況下,如果我們要做一個增加頭像裝飾的小工具,可能需要一個服務器,哪怕沒有人使用,也必須有一臺服務器苦苦支撐,這樣導致有時僅僅是一個Demo,也需要無時無刻的支出成本。但在Serverless架構下,其彈性伸縮特點讓我們不懼怕高併發,其按量付費模式讓我們不懼怕成本支出。