一个难懂的聚簇分类算法

1、抽取全部图像的surf特征（每个图像的特征行不固定，但是列是固定的70）

2、将图像分为两组，一组训练，一组测试

3、将训练图像全部合并为一个大矩阵，并将矩阵聚簇为30个特征。

4、将每一个图像代入聚簇函数，推测每一个图像属于若干个分组（若不够30个分组，后面补1）

5、每个图像就表示为30个特征向量

6、送入逻辑分类进行分类学习

7、得到训练结果

# -*- coding: utf-8 -*-
"""
Created on Thu Aug 11 20:51:19 2016

@author: Administrator
"""



import numpy as np
import mahotas as mh
from mahotas.features import surf
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import *
from sklearn.cluster import MiniBatchKMeans
import glob

#获取文件列表(cat 1,dog 0)
all_instance_filenames = []
all_instance_targets = []
for f in glob.glob('train2/*.jpg'):
    target = 1 if 'cat' in f else 0
    all_instance_filenames.append(f)
    all_instance_targets.append(target)

surf_features = []
counter = 0
for f in all_instance_filenames:
    counter = counter+1
    print 'Reading image:', f,counter/300.0
    image = mh.imread(f, as_grey=True)
    temp_image = surf.surf(image) #[:, 5:] 
    print temp_image.shape
    surf_features.append(temp_image)
    


#分离训练和测试
#分别将训练和测试图像按照行全部罗列起来
train_len = int(len(all_instance_filenames) * .60)
X_train_surf_features = np.concatenate(surf_features[:train_len])

cou1=0
for test1 in surf_features[:train_len]:
    cou1 = cou1+test1.shape[0]
print cou1
print len(X_train_surf_features)

X_test_surf_feautres = np.concatenate(surf_features[train_len:])

y_train = all_instance_targets[:train_len]
y_test = all_instance_targets[train_len:]


n_clusters = 30
print 'Clustering', len(X_train_surf_features), 'features'
estimator = MiniBatchKMeans(n_clusters=n_clusters)
estimator.fit_transform(X_train_surf_features)

    

'''
estimator.cluster_centers_.shape
Out[18]: (30L, 70L)
'''


X_train = []
for instance in surf_features[:train_len]:
    clusters = estimator.predict(instance)
    features = np.bincount(clusters)
    if len(features) < n_clusters:
        features = np.append(features, np.zeros((1, n_clusters-len(features))))
    X_train.append(features)

X_test = []
for instance in surf_features[train_len:]:
    clusters = estimator.predict(instance)
    features = np.bincount(clusters)
    if len(features) < n_clusters:
        features = np.append(features, np.zeros((1, n_clusters-len(features))))
    X_test.append(features)


clf = LogisticRegression(C=0.001, penalty='l2')
clf.fit_transform(X_train, y_train)
predictions = clf.predict(X_test)
print classification_report(y_test, predictions)
print 'Precision: ', precision_score(y_test, predictions)
print 'Recall: ', recall_score(y_test, predictions)
print 'Accuracy: ', accuracy_score(y_test, predictions)