图片预处理

把验证码去噪、二值化、灰度处理

正方教务系统的二维码字符颜色都是蓝色。二维码转为RGB通道，遍历像素点，蓝色改为黑色，其他颜色改为白色。再转为灰度图片，方便后面识别

1
2
3
4
5
6
7
8
9
10
11
12
13
14
def stay_blue2gray(image):
    image = image.convert('RGB')
    height = image.size[0]
    width = image.size[1]
    for y in range(width):
        for x in range(height):
            pix_data = image.getpixel((x, y))
            if pix_data[0] == 0 and pix_data[1] == 0 and pix_data[2] == 153:
                image.putpixel((x, y), (0, 0, 0))
                continue
            else:
                image.putpixel((x, y), (255, 255, 255))
    image = image.convert('L')
    return image

分割图片

验证码的字符总是出现在相同的位置，定长暴力切割

1
2
3
4
5
6
7
8
9
10
def split_image(image):
    images = []
    x = 5
    y = 0
    w = 12
    h = 23
    for i in range(4):
        images.append(image.crop((x, y, x+w, y+h)))
        x += w
    return images

从本地加载字模

字模的文件名切片第一个就是字模对应的字符

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
def ocr(images, model_path='./zfgetcode/data/model'):
    result = ""
    models = []
    file_names = []

    # 加载模型
    for filename in os.listdir(model_path):
        model = Image.open(model_path + "/" + filename)
        file_names.append(filename[0:1])
        models.append(model.convert('L'))

    # 分别识别切割的单子字符
    for image in images:
        result += (single_char_ocr(image, models, file_names))
    return result

OCR识别字符

遍历像素点，返回匹配度最高的结果，匹配算法还是大一程序设计课上，找最小数的穷举算法

avatar

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
def single_char_ocr(image, models, file_names):
    # 识别一个字符
    result = "#"
    height = image.size[0]
    width = image.size[1]
    min_count = width * height
    for i in range(len(models)):
        model = models[i]
        if model.size[1] - width > 2:
            print("OCR_Code.py--71")
            continue
        count = 0
        width_min = width if width < model.size[1] else model.size[1]
        height_min = height if width < model.size[0] else model.size[0]
        for y in range(width_min):
            done = False
            for x in range(height_min):
                if model.getpixel((x, y)) != image.getpixel((x, y)):
                    count += 1
                    if count >= min_count:
                        done = True
                        break
            if done:
                break
        if count <= 3:
            result = file_names[i]
            print(result)
        elif count < min_count:
            min_count = count
            result = file_names[i]
    return result

完整代码(OCR_code.py)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
import os
from PIL import Image
import matplotlib.pyplot as plt


def show_code(image=None, path='./'):
    if image:
        image = Image.open(path + "code.gif")
    plt.figure("CODE")
    plt.imshow(image)
    plt.show()
    return image


def stay_blue2gray(image):
    image = image.convert('RGB')
    height = image.size[0]
    width = image.size[1]
    for y in range(width):
        for x in range(height):
            pix_data = image.getpixel((x, y))
            if pix_data[0] == 0 and pix_data[1] == 0 and pix_data[2] == 153:
                image.putpixel((x, y), (0, 0, 0))
                continue
            else:
                image.putpixel((x, y), (255, 255, 255))
    image = image.convert('L')
    return image


def split_image(image):
    images = []
    x = 5
    y = 0
    w = 12
    h = 23
    for i in range(4):
        images.append(image.crop((x, y, x+w, y+h)))
        x += w
        # images[i].save(str(i) + '.gif')
    return images


def ocr(images, model_path='./zfgetcode/data/model'):
    result = ""
    models = []
    file_names = []

    # 加载模型
    for filename in os.listdir(model_path):
        model = Image.open(model_path + "/" + filename)
        file_names.append(filename[0:1])
        models.append(model.convert('L'))

    # 分别识别切割的单子字符
    for image in images:
        result += (single_char_ocr(image, models, file_names))
    return result


def single_char_ocr(image, models, file_names):
    # 识别一个字符
    result = "#"
    height = image.size[0]
    width = image.size[1]
    min_count = width * height
    for i in range(len(models)):
        model = models[i]
        if model.size[1] - width > 2:
            print("OCR_Code.py--71")
            continue
        count = 0
        width_min = width if width < model.size[1] else model.size[1]
        height_min = height if width < model.size[0] else model.size[0]
        for y in range(width_min):
            done = False
            for x in range(height_min):
                if model.getpixel((x, y)) != image.getpixel((x, y)):
                    count += 1
                    if count >= min_count:
                        done = True
                        break
            if done:
                break
        if count <= 3:
            result = file_names[i]
            print(result)
        elif count < min_count:
            min_count = count
            result = file_names[i]
    return result


def run(image_path='./'):
    image = Image.open(image_path + "code.gif")
    # show_code(image)
    image = stay_blue2gray(image)
    images = split_image(image)
    result = ocr(images)
    # print(result)
    return result


if __name__ == "__main__":
    run()

完整项目包括登录、获取信息和自动选课，其中登录模块的验证码识别采用OCR

OCR字模生成(Java实现)