最近自己做公司系统的UI自动化，结果遇到了前端同事搞得图形验证码，干扰线比较多，PIL识别率很低，因此自己优化了一下，现在识别率从20%提高到了70%左右
处理前：
1.jpg
处理后
test.png
下面是具体代码：
__author__ = "LandiLiu"
from PIL import ImageEnhance, Image
import pytesseract
def getSecurityCode(path):
# 原始图像
image = Image.open (path)
# 亮度增强
enh_bri = ImageEnhance.Brightness (image)
brightness = 1.5
image_brightened = enh_bri.enhance (brightness)
image_brightened.save (path)
# 色度增强
image = Image.open (path)
enh_col = ImageEnhance.Color (image)
color = 1.5
image_colored = enh_col.enhance (color)
image_colored.save (path)
# 对比度增强
image = Image.open (path)
enh_con = ImageEnhance.Contrast (image)
contrast = 1.5
image_contrasted = enh_con.enhance (contrast)
image_contrasted.save (path)
# 锐度增强
image = Image.open (path)
enh_sha = ImageEnhance.Sharpness (image)
sharpness = 3.0
image_sharped = enh_sha.enhance (sharpness)
image_sharped.save (path)
# 黑白化处理
image = Image.open (path)
img_blacked = image.convert ('L')
img_blacked.save (path)
# text = pytesseract.image_to_string (img_blacked)
# print (text)
#去除干扰线
data = Image.open (path)
size = data.size
w = size[0]
h = size[1]
buffer = 0
#进行像素点过滤，判断干扰线像素点8个方向的像素值
try:
for i in range (1, w - 1):
for j in range (1, h - 1):
if data.getpixel ((i, j - 1)) > 150:
buffer += 1
if data.getpixel ((i, j + 1)) > 150:
buffer += 1
if data.getpixel ((i - 1, j)) > 150:
buffer += 1
if data.getpixel ((i + 1, j)) > 150:
buffer += 1
if data.getpixel ((i - 1, j - 1)) > 150:
buffer += 1
if data.getpixel ((i - 1, j + 1)) > 150:
buffer += 1
if data.getpixel ((i + 1, j - 1)) > 150:
buffer += 1
if data.getpixel ((i + 1, j + 1)) > 150:
buffer += 1
# print (buffer)
#对干扰线所在像素点进行置白
if buffer > 4:
data.putpixel ((i, j), 255)
buffer = 0
data.save ('test.png')
text = pytesseract.image_to_string (data)
return text.replace(' ','')
except:return -1
如果有其他的优化方法，欢迎留言
ps：未经作者授权禁止转载
