写python代码，调用opencv库与scipy库，读取文件夹中所有图像，对这些图像进行SIFT特征提取，再对提取到的特征进行mean shift聚类，并得出轮廓系数，将聚类后的图像保存到新的文件夹中

好的，这是一个挺具体的编程要求，下面是我的建议：

你可以按照以下步骤实现这个要求：

1.导入所需的库：

import cv2
import numpy as np
from sklearn.cluster import MeanShift
from scipy import ndimage
import os

2.定义一个函数来实现SIFT特征提取：

def extract_features(image_path, vector_size=32):
    image = cv2.imread(image_path)
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    sift = cv2.xfeatures2d.SIFT_create()
    keypoints, descriptors = sift.detectAndCompute(gray, None)
    if descriptors is None:
        return None
    # 进行特征维度的限制
    if descriptors.size > vector_size:
        descriptors = descriptors[:vector_size]
    return descriptors.flatten()

3.开始进行聚类：

# define your parameters
# 这里的SIFT特征提取生成的特征维度最多为32，可以根据需要进行调整
current_directory = os.getcwd()
image_directory = os.path.join(current_directory, 'input')
output_directory = os.path.join(current_directory, 'output')
vect = np.vectorize(extract_features)
sift_features = vect(image_paths)
sift_features = np.asarray([x for x in sift_features if x is not None])
ms_clf = MeanShift(bandwidth=2)
ms_clf.fit(sift_features)
labels = ms_clf.labels_
cluster_centers = ms_clf.cluster_centers_

labels_unique = np.unique(labels)
n_clusters_ = len(labels_unique)
print("number of estimated clusters : %d" % n_clusters_)

for cluster_id in range(n_clusters_):
    image_indices = np.where(labels == cluster_id)[0]
    print("Image indices for cluster %d : %s" % (cluster_id, str(image_indices)))
    if not os.path.exists(os.path.join(output_directory, str(cluster_id))):
        os.makedirs(os.path.join(output_directory, str(cluster_id)))
    for image_index in image_indices:
        image_path = image_paths[image_index]
        image = cv2.imread(image_path)
        output_path = os.path.join(output_directory, str(cluster_id), os.path.basename(image_path))
        cv2.imwrite(output_path, image)

以上是我给的思路，你可以结合实际情况进行修改，希望对你有所帮助。