Python Model._get_mean_shape方法代码示例

# 需要导入模块: from Model import Model [as 别名]
# 或者: from Model.Model import _get_mean_shape [as 别名]
start_time = time.time()
m1._procrustes_analysis(m1.Patients)
print("--- %s seconds ---" % (time.time() - start_time))
start_time = time.time()
[evals,evecs,num] = m1._PCA(m1.Patients)
print("--- %s seconds ---" % (time.time() - start_time))



m1 = Model()
m1._get_patients(14)
start_time = time.time()
m1._procrustes_analysis(m1.Patients)
print("--- %s seconds ---" % (time.time() - start_time))
mean_shape = m1._get_mean_shape(m1.Patients)
[evals,evecs,num] = m1._PCA(m1.Patients)
tot = sum(evals)
ratio_s = evals / tot
test = np.dot(evecs,_all_teeths.T)
plot(test[:,0])

m1._weight_matrix(m1.Patients)
[a,b]=m1.Patients[1]._alignment_parameters(m1.Patients[2],m1.weight_matrix_)


shape_vectors = np.array([np. ravel(m1.Patients[i].Teeth) for i in range(len(m1.Patients))])
mean = np.mean(shape_vectors, axis=0)
mean = np.reshape(mean, (-1,2))
min_x = min(mean[:,0])
min_y = min(mean[:,1])

# 需要导入模块: from Model import Model [as 别名]
# 或者: from Model.Model import _get_mean_shape [as 别名]
class ModelFitter:
    file_in = 'C:/Users/tangc/Documents/ComVi/_Data/Radiographs/extra/20.tif'
    
    def __init__(self):
        # Loc is the initial location of Active Shape Model
        # Loc can be given by class Locator (automatic initialisation)
        # Loc can also be given by manual initialization (mouse input)
        self.loc = []
        #file_in = image
        self.img = []
        self.model = Model()
        self.model._get_patients(14,8)
    
    def add_graph(self):
        #
        img = cv2.imread(self.file_in)
        gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
        # Clahe parameters were adjusted manually.
        clahe = cv2.createCLAHE(clipLimit=8.0, tileGridSize=(25,25))
        cl1=clahe.apply(gray)
        blur = cv2.GaussianBlur(cl1,(3,3),0)
        blur = cv2.bilateralFilter(blur, 2, 20, 20)
        sobelx = cv2.Sobel(blur,cv2.CV_64F,1,0,ksize=3)
        sobely = cv2.Sobel(blur,cv2.CV_64F,0,1,ksize=3)
        absY = cv2.convertScaleAbs(sobely)
        absX = cv2.convertScaleAbs(sobelx)
        dst = cv2.addWeighted( absX, 0.5, absY, 0.5,0)
        self.img = dst
        
    #def onclick(self,event):
    #    if event.xdata != None and event.ydata != None:
    #        print(event.xdata, event.ydata)
    #        
    #def on_click(self):
    #    im = plt.imread(self.file_in)
    #    self.onclick()
    #    cid = fig.canvas.mpl_connect('button_press_event', self.onclick)
    #    plt.show()
    # 
 
    def init_shape(self):
        init_shape = self.model._get_mean_shape(self.model.Patients)        
        return init_shape
               
    #def __produce_gradient_image(self):
    #    gray = cv2.cvtColor(self.img,cv2.COLOR_BGR2GRAY)
    #    clahe = cv2.createCLAHE(clipLimit=8.0, tileGridSize=(25,25))
    #    cl1=clahe.apply(gray)
    #    blur = cv2.GaussianBlur(cl1,(3,3),0)
    #    blur = cv2.bilateralFilter(blur, 2, 20, 20)
    #    sobelx = cv2.Sobel(blur,cv2.CV_64F,1,0,ksize=3)
    #    sobely = cv2.Sobel(blur,cv2.CV_64F,0,1,ksize=3)       
    #    absX = cv2.convertScaleAbs(sobelx)
    #    absY = cv2.convertScaleAbs(sobely)
    #    dst = cv2.addWeighted( absX, 0.5, absY, 0.5,0)
    #    cv2.imshow('Result', dst)
    #    return dst        
        
    def image_fitting(self,loc,k,ns,loop_num):
        # loop_num is the loop number for fitting process
        img = cv2.imread(self.file_in)
        clone = img.copy()
        cv2.namedWindow("Fitting Window", cv2.WINDOW_AUTOSIZE)
        small = cv2.resize(clone, (0,0), fx=0.3, fy=0.3)
        cv2.imshow( "Fitting Window", small )
        cv2.waitKey(1)
        os.chdir("C:/Users/tangc/Documents/ComVi/ASM")
        shape_datas = pickle.load( open( "save.p", "rb" ) )
        _evals_shape = shape_datas[0]
        _evecs_shape = shape_datas[1]
        #_num_shape = shape_datas[0]
        _grey_mean,_grey_evals,_grey_evecs = self.model.greyscale_PCA(k)
        _init_shape = self.init_shape()
        _centroid = np.mean(_init_shape.Teeth, axis=0)
        _dist_x = _centroid[0] - loc[0]
        _dist_y = _centroid[1] - loc[1]
        _init_shape.Teeth[:,0] = _init_shape.Teeth[:,0] - _dist_x
        _init_shape.Teeth[:,1] = _init_shape.Teeth[:,1] - _dist_y
        init_mean_Teeth = np.copy(_init_shape.Teeth)
        for i in range(loop_num):
            _init_shape.get_normal_to_teeth()
            _Normals = _init_shape.Normals
            l = k + ns
            _init_shape.alter_get_profile_and_Derivatives(l,self.file_in)
            f=np.zeros(shape=(ns*2+1,3200))
            for j in range(3200):
                for n in range(ns*2+1):
                   gi=np.zeros(shape=(2*k+1,1))
                   gi = _init_shape.profiles[n:17+n,j]
                   bc = np.dot(_grey_evecs[j],(gi-_grey_mean[j]))
                   bc = bc/np.sqrt(_grey_evals[j])
                   f[n,j]=np.sum(bc**2)
            temp = np.amin(f, axis=0)
            _idx = np.argmin(f, axis=0)
            movement = (_idx-1)-ns
            _init_shape.Teeth[:,0] = _init_shape.Teeth[:,0] + movement*_Normals[:,0]
            _init_shape.Teeth[:,1] = _init_shape.Teeth[:,1] + movement*_Normals[:,1]
            clone = img.copy()
            for s in range(8):
                #_pointx = _init_shape.Teeth[s,0].astype('int')
#.........这里部分代码省略.........