python图片文字分割 python简单实现图片文字分割
上不了山的小非洲 人气:0想了解python简单实现图片文字分割的相关内容吗，上不了山的小非洲在本文为您仔细讲解python图片文字分割的相关知识和一些Code实例，欢迎阅读和指正，我们先划重点：python图片文字分割,python文字分割,python图片分割，下面大家一起来学习吧。
原图：
图片预处理：图片二值化以及图片降噪处理。
# 图片二值化
def binarization(img,threshold):
#图片二值化操作
width,height=img.size
im_new = img.copy()
for i in range(width):
for j in range(height):
a = img.getpixel((i, j))
aa = 0.30 * a[0] + 0.59 * a[1] + 0.11 * a[2]
if (aa <= threshold):
im_new.putpixel((i, j), (0, 0, 0))
else:
im_new.putpixel((i, j), (255, 255, 255))
# im_new.show() # 显示图像
return im_new
# 图片降噪处理
def clear_noise(img):
# 图片降噪处理
x, y = img.width, img.height
for i in range(x-1):
for j in range(y-1):
if sum_9_region(img, i, j) < 600:
# 改变像素点颜色，白色
img.putpixel((i, j), (255,255,255))
# img = np.array(img)
# # cv2.imwrite('handle_two.png', img)
# # img = Image.open('handle_two.png')
img.show()
return img
# 获取田字格内当前像素点的像素值
def sum_9_region(img, x, y):
"""
田字格
"""
# 获取当前像素点的像素值
a1 = img.getpixel((x - 1, y - 1))[0]
a2 = img.getpixel((x - 1, y))[0]
a3 = img.getpixel((x - 1, y+1 ))[0]
a4 = img.getpixel((x, y - 1))[0]
a5 = img.getpixel((x, y))[0]
a6 = img.getpixel((x, y+1 ))[0]
a7 = img.getpixel((x+1 , y - 1))[0]
a8 = img.getpixel((x+1 , y))[0]
a9 = img.getpixel((x+1 , y+1))[0]
width = img.width
height = img.height
if a5 == 255: # 如果当前点为白色区域,则不统计邻域值
return 2550
if y == 0: # 第一行
if x == 0: # 左上顶点,4邻域
# 中心点旁边3个点
sum_1 = a5 + a6 + a8 + a9
return 4*255 - sum_1
elif x == width - 1: # 右上顶点
sum_2 = a5 + a6 + a2 + a3
return 4*255 - sum_2
else: # 最上非顶点,6邻域
sum_3 = a2 + a3+ a5 + a6 + a8 + a9
return 6*255 - sum_3
elif y == height - 1: # 最下面一行
if x == 0: # 左下顶点
# 中心点旁边3个点
sum_4 = a5 + a8 + a7 + a4
return 4*255 - sum_4
elif x == width - 1: # 右下顶点
sum_5 = a5 + a4 + a2 + a1
return 4*255 - sum_5
else: # 最下非顶点,6邻域
sum_6 = a5+ a2 + a8 + a4 +a1 + a7
return 6*255 - sum_6
else: # y不在边界
if x == 0: # 左边非顶点
sum_7 = a4 + a5 + a6 + a7 + a8 + a9
return 6*255 - sum_7
elif x == width - 1: # 右边非顶点
sum_8 = a4 + a5 + a6 + a1 + a2 + a3
return 6*255 - sum_8
else: # 具备9领域条件的
sum_9 = a1 + a2 + a3 + a4 + a5 + a6 + a7 + a8 + a9
return 9*255 - sum_9
经过二值化和降噪后得到的图片
对图片进行水平投影与垂直投影：
# 传入二值化后的图片进行垂直投影
def vertical(img):
"""传入二值化后的图片进行垂直投影"""
pixdata = img.load()
w,h = img.size
ver_list = []
# 开始投影
for x in range(w):
black = 0
for y in range(h):
if pixdata[x,y][0] == 0:
black += 1
ver_list.append(black)
# 判断边界
l,r = 0,0
flag = False
t=0#判断分割数量
cuts = []
for i,count in enumerate(ver_list):
# 阈值这里为0
if flag is False and count > 0:
l = i
flag = True
if flag and count == 0:
r = i-1
flag = False
cuts.append((l,r))#记录边界点
t += 1
#print(t)
return cuts,t
# 传入二值化后的图片进行水平投影
def horizontal(img):
"""传入二值化后的图片进行水平投影"""
pixdata = img.load()
w,h = img.size
ver_list = []
# 开始投影
for y in range(h):
black = 0
for x in range(w):
if pixdata[x,y][0] == 0:
black += 1
ver_list.append(black)
# 判断边界
l,r = 0,0
flag = False
# 分割区域数
t=0
cuts = []
for i,count in enumerate(ver_list):
# 阈值这里为0
if flag is False and count > 0:
l = i
flag = True
if flag and count == 0:
r = i-1
flag = False
cuts.append((l,r))
t += 1
return cuts,t
这两段代码目的主要是为了分割得到水平和垂直位置的每个字所占的大小，接下来就是对预处理好的图片文字进行分割。
# 创建获得图片路径并处理图片函数
def get_im_path():
OpenFile = tk.Tk()#创建新窗口
OpenFile.withdraw()
file_path = filedialog.askopenfilename()
im = Image.open(file_path)
# 阈值
th = getthreshold(im) - 16
print(th)
# 原图直接二值化
im_new1 = binarization(im, th)
im_new1.show()
# 直方图均衡化
im1 = his_bal(im)
im1.show()
im_new_np = np.array(his_bal(im))
th1 = getthreshold(im1) - 16
print(th1)
# 二值化
im_new = binarization(im1, th1)
# 降噪
im_new_cn = clear_noise(im_new)
height = im_new_cn.size[1]
print(height)
# 算出水平投影和垂直投影的数值
v, vt = vertical(im_new1)
h, ht = horizontal(im_new1)
# 算出分割区域
a = []
for i in range(vt):
a.append((v[i][0], 0, v[i][1], height))
print(a)
im_new.show() # 直方图均衡化后再二值化
# 切割
for i, n in enumerate(a, 1):
temp = im_new_cn.crop(n) # 调用crop函数进行切割
temp.show()
temp.save("c/%s.png" % i)
至此大概就完成了。
接下来是文件的全部代码：
import numpy as np
from PIL import Image
import queue
import matplotlib.pyplot as plt
import tkinter as tk
from tkinter import filedialog#导入文件对话框函数库
window = tk.Tk()
window.title('图片选择界面')
window.geometry('400x100')
var = tk.StringVar()
# 创建获得图片路径并处理图片函数
def get_im_path():
OpenFile = tk.Tk()#创建新窗口
OpenFile.withdraw()
file_path = filedialog.askopenfilename()
im = Image.open(file_path)
# 阈值
th = getthreshold(im) - 16
print(th)
# 原图直接二值化
im_new1 = binarization(im, th)
im_new1.show()
# 直方图均衡化
im1 = his_bal(im)
im1.show()
im_new_np = np.array(his_bal(im))
th1 = getthreshold(im1) - 16
print(th1)
# 二值化
im_new = binarization(im1, th1)
# 降噪
im_new_cn = clear_noise(im_new)
height = im_new_cn.size[1]
print(height)
# 算出水平投影和垂直投影的数值
v, vt = vertical(im_new1)
h, ht = horizontal(im_new1)
# 算出分割区域
a = []
for i in range(vt):
a.append((v[i][0], 0, v[i][1], height))
print(a)
im_new.show() # 直方图均衡化后再二值化
# 切割
for i, n in enumerate(a, 1):
temp = im_new_cn.crop(n) # 调用crop函数进行切割
temp.show()
temp.save("c/%s.png" % i)
# 传入二值化后的图片进行垂直投影
def vertical(img):
"""传入二值化后的图片进行垂直投影"""
pixdata = img.load()
w,h = img.size
ver_list = []
# 开始投影
for x in range(w):
black = 0
for y in range(h):
if pixdata[x,y][0] == 0:
black += 1
ver_list.append(black)
# 判断边界
l,r = 0,0
flag = False
t=0#判断分割数量
cuts = []
for i,count in enumerate(ver_list):
# 阈值这里为0
if flag is False and count > 0:
l = i
flag = True
if flag and count == 0:
r = i-1
flag = False
cuts.append((l,r))#记录边界点
t += 1
#print(t)
return cuts,t
# 传入二值化后的图片进行水平投影
def horizontal(img):
"""传入二值化后的图片进行水平投影"""
pixdata = img.load()
w,h = img.size
ver_list = []
# 开始投影
for y in range(h):
black = 0
for x in range(w):
if pixdata[x,y][0] == 0:
black += 1
ver_list.append(black)
# 判断边界
l,r = 0,0
flag = False
# 分割区域数
t=0
cuts = []
for i,count in enumerate(ver_list):
# 阈值这里为0
if flag is False and count > 0:
l = i
flag = True
if flag and count == 0:
r = i-1
flag = False
cuts.append((l,r))
t += 1
return cuts,t
# 获得阈值算出平均像素
def getthreshold(im):
#获得阈值 算出平均像素
wid, hei = im.size
hist = [0] * 256
th = 0
for i in range(wid):
for j in range(hei):
gray = int(0.3 * im.getpixel((i, j))[0] + 0.59 * im.getpixel((i, j))[1] + 0.11 * im.getpixel((i, j))[2])
th = gray + th
hist[gray] += 1
threshold = int(th/(wid*hei))
return threshold
# 直方图均衡化 提高对比度
def his_bal(im):
#直方图均衡化 提高对比度
# 统计灰度直方图
im_new = im.copy()
wid, hei = im.size
hist = [0] * 256
for i in range(wid):
for j in range(hei):
gray = int(0.3*im.getpixel((i,j))[0]+0.59*im.getpixel((i,j))[1]+0.11*im.getpixel((i,j))[2])
hist[gray] += 1
# 计算累积分布函数
cdf = [0] * 256
for i in range(256):
if i == 0:
cdf[i] = hist[i]
else:
cdf[i] = cdf[i - 1] + hist[i]
# 用累积分布函数计算输出灰度映射函数LUT
new_gray = [0] * 256
for i in range(256):
new_gray[i] = int(cdf[i] / (wid * hei) * 255 + 0.5)
# 遍历原图像，通过LUT逐点计算新图像对应的像素值
for i in range(wid):
for j in range(hei):
gray = int(0.3*im.getpixel((i,j))[0]+0.59*im.getpixel((i,j))[1]+0.11*im.getpixel((i,j))[2])
im_new.putpixel((i, j), new_gray[gray])
return im_new
# 图片二值化
def binarization(img,threshold):
#图片二值化操作
width,height=img.size
im_new = img.copy()
for i in range(width):
for j in range(height):
a = img.getpixel((i, j))
aa = 0.30 * a[0] + 0.59 * a[1] + 0.11 * a[2]
if (aa <= threshold):
im_new.putpixel((i, j), (0, 0, 0))
else:
im_new.putpixel((i, j), (255, 255, 255))
# im_new.show() # 显示图像
return im_new
# 图片降噪处理
def clear_noise(img):
# 图片降噪处理
x, y = img.width, img.height
for i in range(x-1):
for j in range(y-1):
if sum_9_region(img, i, j) < 600:
# 改变像素点颜色，白色
img.putpixel((i, j), (255,255,255))
# img = np.array(img)
# # cv2.imwrite('handle_two.png', img)
# # img = Image.open('handle_two.png')
img.show()
return img
# 获取田字格内当前像素点的像素值
def sum_9_region(img, x, y):
"""
田字格
"""
# 获取当前像素点的像素值
a1 = img.getpixel((x - 1, y - 1))[0]
a2 = img.getpixel((x - 1, y))[0]
a3 = img.getpixel((x - 1, y+1 ))[0]
a4 = img.getpixel((x, y - 1))[0]
a5 = img.getpixel((x, y))[0]
a6 = img.getpixel((x, y+1 ))[0]
a7 = img.getpixel((x+1 , y - 1))[0]
a8 = img.getpixel((x+1 , y))[0]
a9 = img.getpixel((x+1 , y+1))[0]
width = img.width
height = img.height
if a5 == 255: # 如果当前点为白色区域,则不统计邻域值
return 2550
if y == 0: # 第一行
if x == 0: # 左上顶点,4邻域
# 中心点旁边3个点
sum_1 = a5 + a6 + a8 + a9
return 4*255 - sum_1
elif x == width - 1: # 右上顶点
sum_2 = a5 + a6 + a2 + a3
return 4*255 - sum_2
else: # 最上非顶点,6邻域
sum_3 = a2 + a3+ a5 + a6 + a8 + a9
return 6*255 - sum_3
elif y == height - 1: # 最下面一行
if x == 0: # 左下顶点
# 中心点旁边3个点
sum_4 = a5 + a8 + a7 + a4
return 4*255 - sum_4
elif x == width - 1: # 右下顶点
sum_5 = a5 + a4 + a2 + a1
return 4*255 - sum_5
else: # 最下非顶点,6邻域
sum_6 = a5+ a2 + a8 + a4 +a1 + a7
return 6*255 - sum_6
else: # y不在边界
if x == 0: # 左边非顶点
sum_7 = a4 + a5 + a6 + a7 + a8 + a9
return 6*255 - sum_7
elif x == width - 1: # 右边非顶点
sum_8 = a4 + a5 + a6 + a1 + a2 + a3
return 6*255 - sum_8
else: # 具备9领域条件的
sum_9 = a1 + a2 + a3 + a4 + a5 + a6 + a7 + a8 + a9
return 9*255 - sum_9
btn_Open = tk.Button(window,
text='打开图像', # 显示在按钮上的文字
width=15, height=2,
command=get_im_path) # 点击按钮式执行的命令
btn_Open.pack()
# 运行整体窗口
window.mainloop()
加载全部内容