斜视诊断相关尝试

需要用到的库

import pandas as pd
import numpy as np
import cv2
import dlib
from sklearn.manifold import MDS
from sklearn.datasets import load_iris
import matplotlib.pyplot as plt
from sklearn.manifold import Isomap

使用dlib获取人脸上68个特征点

detector = dlib.get_frontal_face_detector()#创建一个容器
predictor = dlib.shape_predictor('shape_predictor_68_face_landmarks.dat')#加载一个自带的分类器
img_path = "2020-3-5-10.jpg"#我需要识别的图片位置
img = cv2.imread(img_path)#使用python-opencv读取图片
img_grey = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)#使图片转化为灰度图片
rects = detector(img_grey, 0)#返回信息
for i in range(len(rects)):
    landmarks = np.matrix([[p.x, p.y] for p in predictor(img_grey,rects[i]).parts()])#获取点的坐标
    for idx, point in enumerate(landmarks):
        # 68点的坐标
        pos = (point[0, 0], point[0, 1])
        #print(idx+1,':',point[0, 0], point[0, 1])
        cv2.circle(img, pos, 2, (255, 0, 0), 1)#画圈圈，（255,0,0）是圈圈的颜色
        font = cv2.FONT_HERSHEY_SIMPLEX
        #cv2.putText(img, str(idx + 1), pos, font, 0.8, (0, 0, 255), 2, cv2.LINE_AA)#为圈圈标上序号
cv2.namedWindow("img", 2)
cv2.imshow("img", img)#展示

cv2.waitKey()
cv2.destroyAllWindows()

处理视频，使用openCV获取视频每一帧

videoCapture = cv2.VideoCapture()
videoCapture.open('1.mp4')

fps = videoCapture.get(cv2.CAP_PROP_FPS)
frames = videoCapture.get(cv2.CAP_PROP_FRAME_COUNT)
#fps是帧率，意思是每一秒刷新图片的数量，frames是一整段视频中总的图片数量。
print("fps=",fps,"frames=",frames)

for i in range(int(frames)):
    ret,frame = videoCapture.read()
    #cv2.imwrite("D:/jupyternotebook/new/1-1.avi(%d).jpg"%i,frame)
    cv2.namedWindow("img", 2)
    cv2.imshow("img", frame)#展示
    cv2.waitKey()
    
cv2.destroyAllWindows()

与提取特征点相结合，提取视频中每一帧的人脸特征点

videoCapture = cv2.VideoCapture()
videoCapture.open('1.mp4')

detector = dlib.get_frontal_face_detector()#创建一个容器
predictor = dlib.shape_predictor('shape_predictor_68_face_landmarks.dat')#加载一个自带的分类器

fps = videoCapture.get(cv2.CAP_PROP_FPS)
frames = videoCapture.get(cv2.CAP_PROP_FRAME_COUNT)
#fps是帧率，意思是每一秒刷新图片的数量，frames是一整段视频中总的图片数量。
print("fps=",fps,"frames=",frames)

for i in range(int(frames)):
    ret,frame = videoCapture.read()
    
    img_grey = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)#使图片转化为灰度图片
    rects = detector(img_grey, 0)#返回信息
    for i in range(len(rects)):
        landmarks = np.matrix([[p.x, p.y] for p in predictor(img_grey,rects[i]).parts()])#获取点的坐标
        for idx, point in enumerate(landmarks):
            # 68点的坐标
            pos = (point[0, 0], point[0, 1])
            cv2.circle(frame, pos, 2, (255, 0, 0), 1)#画圈圈，（255,0,0）是圈圈的颜色
            font = cv2.FONT_HERSHEY_SIMPLEX
            cv2.putText(frame, str(idx + 1), pos, font, 0.8, (0, 0, 255), 2, cv2.LINE_AA)#为圈圈标上序号
    cv2.namedWindow("img", 2)
    cv2.imshow("img", frame)#展示
    cv2.waitKey()
        
cv2.destroyAllWindows()

用来获取特征向量的方法

def getVector(img_path,leftEye=(0,0),rightEye=(0,0)):
    """
    获取关于瞳孔的十二维向量
    """ 
    leftVector,rightVector=[],[] #左右眼的特征向量
    detector = dlib.get_frontal_face_detector()#创建一个容器
    predictor = dlib.shape_predictor('shape_predictor_68_face_landmarks.dat')#加载一个自带的分类器
    img = cv2.imread(img_path)#使用python-opencv读取图片
    img_grey = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)#使图片转化为灰度图片
    rects = detector(img_grey, 0)#返回信息
    
    for i in range(len(rects)):
        landmarks = np.matrix([[p.x, p.y] for p in predictor(img_grey,rects[i]).parts()])#获取点的坐标
        for idx, point in enumerate(landmarks):
            # 68点的坐标
            pos = (point[0, 0], point[0, 1])
            if idx>35 and idx<42:
                leftVector.append(np.sqrt((point[0, 0]-leftEye[0])**2+(point[0, 1]-leftEye[1])**2))
            elif idx>41 and idx<48:
                rightVector.append(np.sqrt((point[0, 0]-rightEye[0])**2+(point[0, 1]-rightEye[1])**2))
            
    # 左右眼眼角距离
    leftMargin=np.sqrt((landmarks[36,0]-landmarks[39,0])**2+(landmarks[36,1]-landmarks[39,1])**2)
    rightMargin=np.sqrt((landmarks[42,0]-landmarks[45,0])**2+(landmarks[42,1]-landmarks[45,1])**2)
    
    ## 由于不同图片会有像素差异，所以需要进行归一化，否则不易分析观察数据
    leftVector=[i/leftMargin for i in leftVector]
    rightVector=[i/rightMargin for i in rightVector]
    
    return leftVector+rightVector #合并为一个十二维的向量

展示正在处理图片的方法，一边处理，一遍展示图片，可以查看特征点提取的效果

def getVectorandDisplay(img_path,leftEye=(0,0),rightEye=(0,0)):
    """
    获取关于瞳孔的十二维向量
    """ 
    leftVector,rightVector=[],[] #左右眼的特征向量
    detector = dlib.get_frontal_face_detector()#创建一个容器
    predictor = dlib.shape_predictor('shape_predictor_68_face_landmarks.dat')#加载一个自带的分类器
    img = cv2.imread(img_path)#使用python-opencv读取图片
    img_grey = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)#使图片转化为灰度图片
    rects = detector(img_grey, 0)#返回信息
    
    for i in range(len(rects)):
        landmarks = np.matrix([[p.x, p.y] for p in predictor(img_grey,rects[i]).parts()])#获取点的坐标
        for idx, point in enumerate(landmarks):
            # 68点的坐标
            pos = (point[0, 0], point[0, 1])
            if idx>35 and idx<42:
                leftVector.append(np.sqrt((point[0, 0]-leftEye[0])**2+(point[0, 1]-leftEye[1])**2))
            elif idx>41 and idx<48:
                rightVector.append(np.sqrt((point[0, 0]-rightEye[0])**2+(point[0, 1]-rightEye[1])**2))
            font = cv2.FONT_HERSHEY_SIMPLEX
            cv2.putText(img, str(idx + 1), pos, font, 0.8, (0, 0, 255), 2, cv2.LINE_AA)#为圈圈标上序号    
            cv2.circle(img, pos, 2, (255, 0, 0), 1)#画圈圈，（255,0,0）是圈圈的颜色
           
    cv2.namedWindow("img", 2)
    cv2.imshow("img", img)#展示
    cv2.waitKey()
    cv2.destroyAllWindows()       
    # 左右眼眼角距离
    leftMargin=np.sqrt((landmarks[36,0]-landmarks[39,0])**2+(landmarks[36,1]-landmarks[39,1])**2)
    rightMargin=np.sqrt((landmarks[42,0]-landmarks[45,0])**2+(landmarks[42,1]-landmarks[45,1])**2)
    
    ## 由于不同图片会有像素差异，所以需要进行归一化，否则不易分析观察数据
    leftVector=[i/leftMargin for i in leftVector]
    rightVector=[i/rightMargin for i in rightVector]
    
    return leftVector+rightVector #合并为一个十二维的向量

将得到的特征向量保存到excel中

def save_data(filePath,data):
    df = pd.DataFrame(data, columns=[i for i in range(0,12)])
    df.to_excel(filePath,index=False)
    print('save success!')

从excel文件中读取文件名和瞳孔位置信息并得到特征向量

data_frame=pd.read_excel('瞳孔中心位置表.xlsx',sheet_name='Sheet1')
#print(len(data_frame['图片名称']))
result=[]
for i in range(len(data_frame['图片名称'])):
    leftEye=(data_frame['左瞳孔位置横坐标'][i],data_frame['左瞳孔位置纵坐标'][i])
    rightEye=(data_frame['右瞳孔位置横坐标'][i],data_frame['右瞳孔位置纵坐标'][i])
    #vector=getVector(data_frame['图片名称'][i],leftEye,rightEye)
    vector=getVectorandDisplay(data_frame['图片名称'][i],leftEye,rightEye)
    result.append(vector)
    print('progress:{}/{}'.format(i+1,len(data_frame['图片名称'])))

save_data("特征向量.xlsx",result)

经过前几步得到特征向量之后，开始进行对数据的降维

通过公式，手动实现的MDS算法

class MyMDS:
    def __init__(self,n_components):
        self.n_components=n_components
    
    def fit(self,data):
        m,n=data.shape
        dist=np.zeros((m,m))
        disti=np.zeros(m)
        distj=np.zeros(m)
        B=np.zeros((m,m))
        for i in range(m):
            dist[i]=np.sum(np.square(data[i]-data),axis=1).reshape(1,m)
        for i in range(m):
            disti[i]=np.mean(dist[i,:])
            distj[i]=np.mean(dist[:,i])
        distij=np.mean(dist)
        for i in range(m):
            for j in range(m):
                B[i,j] = -0.5*(dist[i,j] - disti[i] - distj[j] + distij)
        lamda,V=np.linalg.eigh(B)
        index=np.argsort(-lamda)[:self.n_components]
        diag_lamda=np.sqrt(np.diag(-np.sort(-lamda)[:self.n_components]))
        V_selected=V[:,index]
        Z=V_selected.dot(diag_lamda)
        return Z

对数据降维并作图

data_frame1=pd.read_excel('D:\\Study\\毕业设计\\数据集\\正常眼位\\特征向量.xlsx',sheet_name='Sheet1')
size1=25
data_frame2=pd.read_excel('D:\\Study\\毕业设计\\数据集\\假性内斜视（正常眼位）\\特征向量.xlsx',sheet_name='Sheet1')
size2=6
data_frame3=pd.read_excel('D:\\Study\\毕业设计\\数据集\\内斜视\\特征向量.xlsx',sheet_name='Sheet1')
size3=5
data_frame4=pd.read_excel('D:\\Study\\毕业设计\\数据集\\外斜视\\特征向量.xlsx',sheet_name='Sheet1')
size4=2
data_frame5=pd.read_excel('D:\\Study\\毕业设计\\数据集\\斜视视频\\特征向量.xlsx',sheet_name='Sheet1')
size5=4;
data_frame6=pd.read_excel('D:\\Study\\毕业设计\\数据集\\正常人眼\\特征向量.xlsx',sheet_name='Sheet1')
size6=40;


data=np.vstack((np.array(data_frame6),np.array(data_frame1),np.array(data_frame2),np.array(data_frame3),np.array(data_frame4),np.array(data_frame5)))

# 用sklearn库
# clf=MDS(2)
# clf.fit(data)
# eyes=clf2.fit_transform(data)

# 用手写的库
clf=MyMDS(2)
eyes=clf.fit(data)

#plt.scatter(eyes[:,0],eyes[:,1],c=np.hstack((np.zeros(size6),np.ones(size1),np.ones(size2)*2,np.ones(size3)*3,np.ones(size4)*4,np.ones(size5)*5)))
plt.scatter(eyes[:,0],eyes[:,1],c=np.hstack((np.zeros(size6),np.ones(size1+size2+size3+size4+size5))))
plt.title('Using sklearn MDS')

使用sklearn的mds降到二维

使用按照公式的mds降到二维

使用matlab对mds降到三维的数据进行可视化

使用流形进行降维

iso=Isomap(n_components=2)
iso.fit(data)
eyes=iso.transform(data)
plt.scatter(eyes[:,0],eyes[:,1],c=np.hstack((np.zeros(size6),np.ones(size1+size2+size3+size4+size5))))
plt.title('Using sklearn Isomap')

使用sklearn的ISOMAP流形降到二维

使用matlab对流形降到三维的数据进行可视化

使用PCA进行降维

from sklearn.decomposition import PCA
pca=PCA(n_components=2)
pca.fit(data)
eyes=pca.fit_transform(data)
plt.scatter(eyes[:,0],eyes[:,1],c=np.hstack((np.zeros(size6),np.ones(size1+size2+size3+size4+size5))))
plt.title('Using sklearn PCA')

使用sklearn的PCA降到二维

使用matlab对PCA降到三维的数据进行可视化

处理后的数据

1.jpg

[154.00324671902212, 68.6804193347711, 55.31726674375732, 140.42791745233566, 83.23460818673925, 101.40019723846694] 
[0.5404953715425003, 0.24104328679361975, 0.19414348254574618, 0.49285090435021756, 0.29212319503345424, 0.3558778041925428] 

[118.33849753989612, 40.45985664828782, 90.47098982546837, 185.13238506539042, 118.70130580579136, 76.4198926981712]
[0.40547336548893476, 0.13863108441824074, 0.3099885285536982, 0.6343350033504502, 0.4067164866342759, 0.26184404675396683]

瞳孔坐标：

leftEye,rightEye=(1215,1587),(1923,1495)

眼角间距

left=284.92981591964013
right=291.8527025744322

左眼六个特征点：

37 : 1069 1636
38 : 1149 1568
39 : 1257 1551
40 : 1353 1613
41 : 1263 1655
42 : 1154 1668

右眼六个特征点：

43 : 1821 1555
44 : 1892 1469
45 : 2006 1459
46 : 2108 1502
47 : 2026 1554
48 : 1915 1571

2.jpg

[262.00763347658403, 161.533278305122, 69.18092222571191, 118.72657663724664, 91.78235124467012, 179.2344832893492] 
[0.7120403619914214, 0.4389880265390032, 0.18800829674643307, 0.32265515887857754, 0.24943066634170935, 0.4870933898729304] 

[142.6954799564443, 58.52349955359813, 135.988970140964, 233.93161393877486, 160.0312469488381, 95.63472172804185]
[0.3938289380700953, 0.1615205168963613, 0.37531946853683834, 0.6456338990320662, 0.4416743688246573, 0.2639447368079219]

瞳孔坐标：

leftEye,rightEye=(1241,1581),(1969,1605)

眼角间距

left=367.96738985948195
right=362.3285801589491

左眼六个特征点：

37 : 979 1579
38 : 1099 1504
39 : 1246 1512
40 : 1341 1645
41 : 1223 1671
42 : 1076 1651

右眼六个特征点：

43 : 1840 1666
44 : 1949 1550
45 : 2092 1547
46 : 2201 1635
47 : 2102 1694
48 : 1958 1700

3.jpg

[22.02271554554524, 13.0, 3.1622776601683795, 11.313708498984761, 10.198039027185569, 16.1245154965971] 
[0.7148875982891058, 0.4219978575547771, 0.10265187673726074, 0.36725851904391094, 0.3310423554409472, 0.5234239225902136] 

[9.899494936611665, 3.1622776601683795, 15.297058540778355, 26.1725046566048, 17.88854381999832, 10.44030650891055]
[0.2978049408680619, 0.09513029883089882, 0.46017899330842227, 0.7873433191832159, 0.53813823519705, 0.31407408988430036]

瞳孔坐标：

leftEye,rightEye=(60,116),(122,119)

眼角间距

left=30.805843601498726 
right=33.24154027718932

左眼六个特征点：

37 : 38 117
38 : 48 111
39 : 61 113
40 : 68 124
41 : 58 126
42 : 46 124

右眼六个特征点：

43 : 115 126
44 : 123 116
45 : 137 116
46 : 148 122
47 : 138 127
48 : 125 129

降到三维的数据,使用matlab画图看看

画图命令

>> scatter3(data(1:40,1),data(1:40,2),data(1:40,3),'r','filled')
>> hold on
>> scatter3(data(40:82,1),data(40:82,2),data(40:82,3),'b','filled')

用mds的数据

[[-2.49399757e-01  1.45448453e-01  3.14430200e-01]
 [ 3.10091160e-02  1.53125557e-01  4.07341840e-02]
 [-1.88586289e-02  2.18464862e-01  1.97185188e-01]
 [ 6.86836401e-02  1.36796820e-01 -1.87589551e-01]
 [ 2.77593027e-01 -1.61774982e-01 -2.86602565e-01]
 [ 1.93983048e-01  1.92304257e-01 -3.13370198e-01]
 [ 1.27267756e-01 -7.67659896e-02 -4.52851354e-02]
 [-3.00611225e-02  1.63941239e-01 -3.36880102e-03]
 [-6.22824906e-02  1.32722780e-01  1.23812752e-01]
 [ 1.08006730e-02 -1.01588882e-01  2.31511499e-02]
 [ 3.30004053e-02 -5.20796132e-02 -2.32344974e-02]
 [-1.31601481e-01  8.29467510e-02 -1.43420491e-01]
 [ 7.09714573e-02  5.86236382e-03  1.09234383e-01]
 [-4.90822001e-01  8.49045714e-02  5.55708935e-01]
 [ 3.18659410e-01 -1.08741706e-01 -4.29548153e-01]
 [-9.27353852e-02  1.99098281e-01 -2.09444243e-01]
 [ 1.34360044e-01  1.89308097e-02  1.80100627e-01]
 [-4.97027359e-01  1.62379395e-01  4.83063348e-01]
 [ 3.40908669e-01 -2.75238568e-02 -4.23000292e-01]
 [-1.83064599e-01  2.06038642e-01  1.82165215e-01]
 [ 2.51228500e-01  2.98243289e-03  1.21059295e-01]
 [ 8.41723581e-02  3.07782913e-02  9.29748861e-02]
 [ 3.23868293e-01  2.17797782e-02 -3.53623052e-01]
 [-4.55957829e-01  1.05119658e-01  4.94820768e-01]
 [ 3.63650091e-01  1.13752464e-01 -2.01811256e-01]
 [-8.83835306e-02  1.83714631e-01  4.15943063e-01]
 [ 1.36501334e-01  1.75586390e-01  9.49316099e-03]
 [ 1.13098846e-01  1.64817407e-01  8.92369850e-02]
 [ 4.39526042e-01  5.51961444e-02 -4.08340788e-01]
 [-3.14714054e-01  1.60718539e-01  5.32234194e-01]
 [ 2.31937359e-02  1.90351665e-01  1.71290778e-01]
 [ 2.59390470e-01  5.34437674e-02  1.52806729e-01]
 [ 2.28086933e-01  1.05735501e-01 -5.73952252e-02]
 [ 4.84280966e-01 -1.59859399e-02 -5.35017086e-01]
 [-1.57558147e-01  1.22139474e-01  4.87683144e-01]
 [ 4.89499178e-02  1.15521286e-01  2.01601157e-01]
 [-2.01947203e-01  3.97304916e-02  1.07735921e-02]
 [-8.78445428e-03  1.73018151e-01  5.93306139e-02]
 [-6.34135328e-02  2.42497499e-01  2.32621892e-01]
 [ 1.86815757e-01  2.40863576e-01 -8.08242379e-02]
 [ 3.18872412e-02 -1.65928475e-01  7.24810266e-02]
 [ 4.67124151e-02 -1.23505811e-01 -6.81669937e-02]
 [-8.58688090e-02 -1.75471927e-01  4.04126955e-02]
 [ 2.50412135e-02 -5.84173306e-03 -6.38159360e-02]
 [ 4.45272382e-04 -9.55700418e-02 -4.17753037e-02]
 [ 7.68480180e-02 -2.03036729e-01 -3.25003115e-02]
 [-5.13683882e-02  1.30150314e-02  1.27008183e-02]
 [-1.19095616e-01  2.63935845e-02  1.67217121e-01]
 [-1.55731569e-01 -6.13164643e-02 -5.79288461e-02]
 [ 2.55787251e-02  1.54398406e-01 -4.43651283e-02]
 [-5.81074023e-02 -5.96704110e-02 -1.92004874e-02]
 [-1.66512406e-01  1.74282574e-01  6.09214145e-02]
 [ 6.28739918e-03 -1.98723992e-01 -1.04908191e-01]
 [-8.15111783e-02 -2.29574457e-02  1.73317682e-02]
 [ 1.56155364e-01 -1.36447088e-01  1.68105432e-02]
 [ 7.53911245e-02  6.32822576e-02 -9.99458715e-02]
 [-5.99027304e-02 -1.50068789e-01 -9.64137979e-03]
 [ 8.24694650e-03 -3.91661560e-02  2.51559303e-02]
 [-1.01531932e-02 -3.77815004e-02 -4.99666618e-03]
 [-5.35054474e-02 -7.30392125e-02  8.76396443e-02]
 [ 1.37050875e-01 -3.77276138e-02 -1.41287900e-01]
 [-1.08268466e-01 -1.26201413e-01 -9.08894233e-03]
 [-7.10319703e-02 -3.04260893e-01  1.14380947e-01]
 [-1.04416373e-02 -4.30733311e-02  1.10783214e-02]
 [ 8.05725900e-02  1.00358944e-01 -7.19735073e-02]
 [-1.77939595e-01  3.78657096e-01 -5.82221823e-01]
 [-6.27098576e-02 -2.50191890e-01  7.97481714e-02]
 [ 5.84417931e-03 -9.89380078e-02 -1.61896697e-02]
 [-9.34074536e-02 -3.30539259e-03  5.02316859e-02]
 [-5.28025442e-03 -2.36356795e-01  2.72266260e-01]
 [ 7.50397456e-02 -1.01385009e-01  2.79782957e-02]
 [-1.48321924e-01 -6.44991629e-01 -7.80080191e-02]
 [-1.11741020e-02 -6.15989830e-01 -1.50236678e-01]
 [-8.38360269e-01 -2.65106533e-02  8.05583241e-02]
 [-2.91953568e-03 -2.88930589e-01 -8.97861457e-02]
 [-1.87777170e-01 -4.03518537e-01 -1.18786983e-02]
 [ 1.98221038e-01  1.06606067e-01 -1.78341096e-01]
 [-3.43053807e-02  1.05110180e-02  9.56699036e-02]
 [ 2.28586840e-01 -5.00239012e-01 -2.63835005e-01]
 [ 5.06795497e-02  2.62819320e-01 -1.48286343e-01]
 [ 9.95341103e-02 -6.85937008e-02  5.28356041e-02]
 [-2.37817208e-01  3.52164814e-01 -5.76620203e-01]]

用流型的数据

[[-0.52475938 -0.20655011 -0.08751006]
 [-0.04211118 -0.28677996  0.01283477]
 [-0.23290125 -0.31975843 -0.02640291]
 [ 0.20882962 -0.17784644  0.11685629]
 [ 0.48612152  0.04415368 -0.20709801]
 [ 0.40810652 -0.29761106  0.06016139]
 [ 0.11023122  0.17601449  0.03529748]
 [ 0.005006   -0.24880078  0.06481727]
 [-0.14672938 -0.20884888  0.00231434]
 [-0.00136789  0.18036852  0.05904796]
 [ 0.04346648  0.13484799  0.0517976 ]
 [-0.2086759   0.01539688  0.03089669]
 [-0.02166921  0.05615281  0.05751615]
 [-0.9323577  -0.28793163 -0.14319581]
 [ 0.7438372  -0.06377497 -0.32139509]
 [ 0.15989466 -0.11604386  0.30746877]
 [-0.06454647 -0.06882085 -0.09234102]
 [-0.86085057 -0.29155556 -0.13762266]
 [ 0.69298832 -0.2106876  -0.30563432]
 [-0.2975549  -0.19136725 -0.00841271]
 [ 0.07038077 -0.12780482 -0.11184737]
 [-0.02744121  0.02193468  0.03589945]
 [ 0.6181052  -0.29953859 -0.24259442]
 [-0.86620269 -0.28489955 -0.14038095]
 [ 0.40462772 -0.30743304 -0.09766412]
 [-0.4880671  -0.4144327  -0.08075041]
 [ 0.07753252 -0.27100687 -0.02250921]
 [-0.02967027 -0.20548343 -0.04672175]
 [ 0.64824562 -0.3104454  -0.32720432]
 [-0.79625244 -0.29221978 -0.13054429]
 [-0.1708873  -0.28771935 -0.08488609]
 [-0.01408447 -0.16517332 -0.11130276]
 [ 0.20009263 -0.22473314 -0.07211356]
 [ 0.81381735 -0.29720148 -0.43431405]
 [-0.63170911 -0.41809868 -0.14217846]
 [-0.14625525 -0.24858571 -0.08234401]
 [-0.23634134  0.06116928  0.04892996]
 [-0.12169358 -0.27794344  0.01480048]
 [-0.30445435 -0.32216279 -0.06952904]
 [ 0.19751155 -0.3148039   0.04824583]
 [-0.01033803  0.2991925   0.07438641]
 [ 0.11639365  0.19090486  0.08385814]
 [ 0.00803306  0.2946195   0.12960284]
 [ 0.08978154  0.04896987  0.08392962]
 [ 0.01349591  0.18915705  0.04375113]
 [ 0.05596794  0.30491331  0.08185548]
 [-0.01238723  0.10574221  0.08974813]
 [-0.17636472  0.01132155  0.04554243]
 [-0.13887366  0.20196536  0.06782757]
 [ 0.05540129 -0.1217556   0.11746257]
 [ 0.02419252  0.1430622   0.08017393]
 [-0.23199722 -0.11973721 -0.02459098]
 [ 0.03850406  0.29769732  0.02310231]
 [-0.11678057  0.1203215   0.03562951]
 [ 0.02983969  0.23687132  0.10040883]
 [ 0.13568508 -0.07715057  0.0646312 ]
 [ 0.01909661  0.23559501  0.09642388]
 [-0.00676058  0.08938025  0.05692723]
 [ 0.00422334  0.09796682  0.04683499]
 [-0.02535596  0.09221617  0.09413002]
 [ 0.18839983  0.06549906  0.05666795]
 [-0.11109553  0.21801478  0.04193411]
 [-0.12927412  0.45430992  0.10660688]
 [ 0.01498223  0.10207633  0.05911291]
 [ 0.09477913 -0.18507621  0.03382207]
 [ 0.58050952 -0.35921256  0.59887673]
 [-0.01366191  0.3867615   0.13295155]
 [-0.01234158  0.18022319  0.03386664]
 [-0.1450861   0.0811554   0.03249574]
 [-0.0959389   0.46617629  0.16799436]
 [ 0.03331177  0.21424554  0.06412596]
 [ 0.19085693  0.77608487 -0.26830046]
 [ 0.275868    0.68862416 -0.34028163]
 [-0.95682389  0.73859395 -0.12494235]
 [ 0.04485674  0.39321044 -0.04296051]
 [ 0.01679716  0.61337447  0.10119243]
 [ 0.24806421 -0.20181272 -0.01433096]
 [-0.08224569  0.06483277  0.06079385]
 [ 0.43870353  0.4925348  -0.44040973]
 [ 0.24468503 -0.22743142  0.18590908]
 [ 0.02641446  0.10096763  0.10855583]
 [ 0.55427047 -0.34838057  0.63429734]]

使用pca得到的数据

[[ 0.39873983 -0.0954298  -0.01807273]
 [ 0.02899003 -0.14997171 -0.03957421]
 [ 0.18471675 -0.19573283 -0.08449226]
 [-0.16238917 -0.14918477  0.06989039]
 [-0.35487308  0.06213577  0.04098967]
 [-0.33337612 -0.22360492  0.08426196]
 [-0.11859844  0.05511917 -0.06692357]
 [ 0.03270696 -0.14437857  0.0175223 ]
 [ 0.14513263 -0.11564825 -0.01946778]
 [-0.00152372  0.09521567 -0.02486575]
 [-0.0406502   0.04572753 -0.01302994]
 [-0.02357004 -0.06403917  0.05712957]
 [ 0.03809298 -0.01896027 -0.09714492]
 [ 0.72321432 -0.01622871  0.11513574]
 [-0.5385383   0.04002175 -0.01414173]
 [-0.06790162 -0.16885438  0.18257616]
 [ 0.05480324 -0.03349948 -0.2035435 ]
 [ 0.68884266 -0.094559    0.08580841]
 [-0.53875304 -0.03667426 -0.00686107]
 [ 0.26136117 -0.14300241  0.04771036]
 [-0.06146883 -0.03506913 -0.24607944]
 [ 0.02373742 -0.0414455  -0.1087627 ]
 [-0.47084935 -0.07252825 -0.01783601]
 [ 0.66887559 -0.05028923  0.08689585]
 [-0.36499023 -0.15538154 -0.10602449]
 [ 0.39897757 -0.14861073 -0.1586591 ]
 [-0.05344611 -0.18937034 -0.07579399]
 [ 0.01870045 -0.171886   -0.1244774 ]
 [-0.5727496  -0.11714474 -0.0888422 ]
 [ 0.62025387 -0.09816015 -0.08050394]
 [ 0.14017167 -0.17901554 -0.09968552]
 [-0.03511426 -0.07397216 -0.25708497]
 [-0.16759328 -0.13576774 -0.09841746]
 [-0.69672747 -0.0729742  -0.03584706]
 [ 0.49191761 -0.0835073  -0.15570585]
 [ 0.13785095 -0.11294631 -0.15383945]
 [ 0.12612221 -0.00476644  0.10914945]
 [ 0.06078638 -0.13925049 -0.04350776]
 [ 0.24146364 -0.2074467  -0.07032349]
 [-0.14805162 -0.26493629 -0.06574267]
 [ 0.01049939  0.15533315 -0.06887585]
 [-0.07746581  0.09421187 -0.00459785]
 [ 0.05891246  0.17318522  0.03958503]
 [-0.05706317 -0.00302849  0.01889711]
 [-0.04105077  0.09122783  0.01306263]
 [-0.09664847  0.17824186 -0.04854476]
 [ 0.04403064 -0.00205085  0.042824  ]
 [ 0.19599879 -0.00560674 -0.01008214]
 [ 0.03210921  0.07655803 -0.00086202]
 [-0.02351106 -0.14586059  0.02526092]
 [ 0.02003864  0.06096351  0.04713513]
 [ 0.15384366 -0.1219016   0.02379801]
 [-0.10401179  0.18757312  0.03517435]
 [ 0.04847323  0.03902306  0.00878325]
 [-0.08934269  0.10156385 -0.0987567 ]
 [-0.11143968 -0.07779589  0.01144192]
 [ 0.01787757  0.11936668  0.02537317]
 [ 0.01160642  0.036014   -0.01284288]
 [ 0.00075838  0.03689241  0.01512577]
 [ 0.07266308  0.06445297  0.01339698]
 [-0.19782268  0.01145398 -0.02139167]
 [ 0.03430166  0.13709737  0.01841701]
 [ 0.08075269  0.30092799 -0.02705681]
 [ 0.00918671  0.04137362  0.00908525]
 [-0.08734214 -0.10612491 -0.01628773]
 [-0.26909675 -0.36755566  0.54610944]
 [ 0.06516807  0.2458186  -0.01757508]
 [-0.02725842  0.09263712 -0.00644623]
 [ 0.08575142  0.0226102   0.01696553]
 [ 0.17235365  0.23380595 -0.16298657]
 [-0.03840252  0.08655714 -0.07854006]
 [-0.05806153  0.64913937  0.07023196]
 [-0.16766007  0.56698742  0.11545025]
 [ 0.54819843  0.31192379  0.46126256]
 [-0.09962754  0.27629622  0.02916976]
 [ 0.05232281  0.40652743  0.10475996]
 [-0.24291227 -0.13861722 -0.02620219]
 [ 0.09531742 -0.0034726  -0.02076486]
 [-0.39345684  0.41950536  0.01389701]
 [-0.10757596 -0.25800411  0.08943682]
 [-0.02253755  0.04579144 -0.07990071]
 [-0.23217009 -0.3270245   0.58525142]]