基本思路
- 获取图像
- 对图像进行二值化处理
- 识别图中轮廓
- 识别轮廓文字
- 构建待选文字图像
- 比较识别文字与待选文字图像
- 返回点选结果
本文仅提供基本思路，具体应用需根据各类型点选验证码自行修改。
实现过程
0.导入依赖
import base64 import json import cv2 import numpy as np import requests from PIL import Image, ImageDraw, ImageFont
- 获取图像
第一步是获取图像，有一些验证码图像是以二进制形式返回的，本文测试的图像是以base64编码字符串的形式返回，因此需要对其进行解码。
def getData(capsession: requests.Session): resp = s.post("验证码获取url") return resp.json()["repData"] def getImageFromBase64(b64): buffer = base64.b64decode(b64) nparr = np.frombuffer(buffer, np.uint8) image = cv2.imdecode(nparr, cv2.IMREAD_COLOR) return image
2.对图像进行二值化处理
为了方便识别文字轮廓，我们对图像进行二值化处理。
def normalizeImage(img): img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) _, img = cv2.threshold(img, 1, 255, cv2.THRESH_BINARY) img = cv2.bitwise_not(img) return img
3.识别图中轮廓
在识别图中轮廓时，为了提高效率和准确度，我们对轮廓按长宽比和面积进行筛选，尽可能保证轮廓能够满足文字识别需要。
def findContour(img): contours, _ = cv2.findContours(img, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE) def find_if_close(cnt1, cnt2): row1, row2 = cnt1.shape[0], cnt2.shape[0] for i in range(row1): for j in range(row2): dist = np.linalg.norm(cnt1[i] - cnt2[j]) if abs(dist) < 5: returnTrue elif i == row1 - 1and j == row2 - 1: returnFalse LENGTH = len(contours) status = np.zeros((LENGTH, 1)) for i, cnt1 in enumerate(contours): x = i if i != LENGTH - 1: for j, cnt2 in enumerate(contours[i + 1:]): x = x + 1 dist = find_if_close(cnt1, cnt2) if dist == True: val = min(status[i], status[x]) status[x] = status[i] = val else: if status[x] == status[i]: status[x] = i + 1 unified = [] maximum = int(status.max()) + 1 for i in range(maximum): pos = np.where(status == i)[0] if pos.size != 0: cont = np.vstack([contours[i] for i in pos]) hull = cv2.convexHull(cont) unified.append(hull) cnt = list(filter(aspectRatio,unified)) cnt.sort(key=cv2.contourArea,reverse=True) return cnt[:4] def aspectRatio(cnt): _,_,w,h = cv2.boundingRect(cnt) return (0.6<float(w)/h<1.7) and (cv2.contourArea(cnt)>200.0)
4.识别轮廓文字
我们对图中文字轮廓进行识别，返回文字轮廓与相应的坐标位置。
def extractCharContour(img, contour): mult = 1.2 ret = [] point = [] for cnt in contour: rect = cv2.minAreaRect(cnt) box = cv2.boxPoints(rect) box = np.int0(box) W = rect[1][0] H = rect[1][1] Xs = [i[0] for i in box] Ys = [i[1] for i in box] x1 = min(Xs) x2 = max(Xs) y1 = min(Ys) y2 = max(Ys) rotated = False angle = rect[2] if angle < -45: angle += 90 rotated = True center = (int((x1 + x2) / 2), int((y1 + y2) / 2)) size = (int(mult * (x2 - x1)), int(mult * (y2 - y1))) try: M = cv2.getRotationMatrix2D((size[0] / 2, size[1] / 2), angle, 1.0) cropped = cv2.getRectSubPix(img, size, center) cropped = cv2.warpAffine(cropped, M, size) croppedW = W ifnot rotated else H croppedH = H ifnot rotated else W croppedRotated = cv2.getRectSubPix( cropped, (int(croppedW * mult), int(croppedH * mult)), (size[0] / 2, size[1] / 2), ) im = cv2.resize(croppedRotated, (20, 20)) kernel = np.array([[0, -1, 0], [-1, 5, -1], [0, -1, 0]], np.float32) im = cv2.filter2D(im, -1, kernel=kernel) ret.append(im) point.append((rect[0][0], rect[0][1])) except: pass return ret, point
5.构建待选文字图像
将获取的待选文字转化为图像。
def genCharacter(ch, size): img = Image.new("L", size, 0) font = ImageFont.truetype("simsun.ttc", min(size)) draw = ImageDraw.Draw(img) draw.text((0, 0), ch, font=font, fill=255) return np.asarray(img)
6.比较识别文字与待选文字图像
我们将识别文字轮廓与待选文字图像进行比较，获得对应位置，并根据验证码要求依次添加点选坐标。
def compareCharImage(words, chars, point, word_list) scores = [] for i, word in enumerate(words): for j, char in enumerate(chars): scores.append(((i, j), cv2.bitwise_xor(char, word).sum())) scores.sort(key=lambda x: x[1]) word_set = set() char_set = set() answers = {} for score in scores: if (score [0][0] notin word_set) and (score [0][1] notin char_set): continue word_set.add(score[0][0]) char_set.add(score[0][1]) answers[word_list[score[0][0]]] = point[score[0][1]] return [{ "x": int(answers[word][0]), "y": int(answers[word][1]) } for word in word_list]
7.返回点选结果
最后，将得到的点选结果返回给服务器进行验证。
def checkCaptcha(captchaSession: requests.Session, data, point): enc = encrypt(json.dumps(point).replace(" ", ""), data["secretKey"]) resp = captchaSession.post( "验证码提交url", json={ "token": data["token"], "pointJson": enc, }, ) return resp.json()
