SVM 算法处理手写识别体（包含如何处理原始图片的代码）

github 地址，包含数据集。
https://github.com/ranran4082391/ran_11
# coding=gbk
from PIL import Image
import numpy as np
import os
from sklearn.datasets import load_digits  # 加载手写数字识别数据
from sklearn.model_selection import train_test_split  # 训练测试数据分割
from sklearn.preprocessing import StandardScaler  # 标准化工具
from sklearn.svm import LinearSVC
from sklearn.metrics import classification_report  # 预测结果分析工具


def ImageToMatrix(filename):
    # 读取图片
    im = Image.open(filename)
    img = im.resize((32, 32), Image.ANTIALIAS)
    img = img.convert("1")
    data = img.getdata()
    data = np.matrix(data, dtype='float')/255.0
    return np.array(data)

mat = []
for filename in os.walk('train'):
    for list_f in filename[2]:
        data = ImageToMatrix('train/'+list_f)
        print(data)
        mat.append(data.reshape(1, -1).tolist()[0])

X = np.array(mat)
y = np.array([x for x in range(0, 10)])


test_filename = r'train\9.jpg'

test_data = ImageToMatrix(test_filename)
print(test_data[0])
svm = LinearSVC()
svm.fit(X, y)
print(svm.predict(test_data))

'''
以上是解决问题的基本方法，下面是包含大量训练集与测试集的样本，更有训练价值。
'''
digits = load_digits()
# 数据纬度，1797幅图，8*8
print(digits.data.shape)

# 分割数据
X_train, X_test, Y_train, Y_test = train_test_split(digits.data, digits.target, test_size=0.25, random_state=33)

ss = StandardScaler()
# fit是实例方法，必须由实例调用
X_train = ss.fit_transform(X_train)
X_test = ss.transform(X_test)

lsvc = LinearSVC()
lsvc.fit(X_train, Y_train)

Y_predict = lsvc.predict(X_test)

print(classification_report(Y_test, Y_predict, target_names=digits.target_names.astype(str)))