SIFT+SVM图像分类

import os
import numpy as np
import cv2
import glob
import sklearn.svm as svm
import joblib

def calcSiftFeature(img):
    #设置图像sift特征关键点最大为200
    sift = cv2.xfeatures2d.SURF_create()
    #计算图片的特征点和特征点描述
    keypoints, features = sift.detectAndCompute(img, None)
    return features

#计算词袋
def learnVocabulary(features):
    wordCnt = 50
    #criteria表示迭代停止的模式   eps---精度0.1，max_iter---满足超过最大迭代次数20
    criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 20, 0.1)
    #得到k-means聚类的初始中心点
    flags = cv2.KMEANS_RANDOM_CENTERS
    # 标签，中心 = kmeans(输入数据（特征)、聚类的个数K,预设标签，聚类停止条件、重复聚类次数、初始聚类中心点
    compactness, labels, centers = cv2.kmeans(features, wordCnt, None,criteria, 20, flags)
    return centers

#计算特征向量
def calcFeatVec(features, centers):
    featVec = np.zeros((1, 50))
    for i in range(0, features.shape[0]):
        <