Descripteurs de Texture : Gray Level Co-occurence Matrix (GLCM) et Local Binary Pattern (LBP) скачать в хорошем качестве

Descripteurs de Texture : Gray Level Co-occurence Matrix (GLCM) et Local Binary Pattern (LBP)

-*- coding: utf-8 -*- import numpy as np import cv2 import matplotlib.pyplot as plt from skimage import feature #%% Load image image_name = 'fge2.jpg' A_bgr = cv2.imread(image_name) A_rgb = cv2.cvtColor(A_bgr, cv2.COLOR_BGR2RGB); image = cv2.cvtColor(A_bgr, cv2.COLOR_BGR2GRAY) plt.close('all'); #%% GLCM Features extraction of ROI chosen manually T = 80 ROI = np.empty((7,T,T),np.uint8); mask = np.ones((T,T)) x,y=1100,410 ; ROI[0,:,:] = image[y:y+T,x:x+T]; A_rgb_ROI = cv2.rectangle(A_rgb, (x,y), (x+T,y+T), 255,4) x,y=1000,1400 ; ROI[1,:,:] = image[y:y+T,x:x+T]; A_rgb_ROI = cv2.rectangle(A_rgb_ROI, (x,y), (x+T,y+T), 255,4) x,y=800,1000 ; ROI[2,:,:] = image[y:y+T,x:x+T]; A_rgb_ROI = cv2.rectangle(A_rgb_ROI, (x,y), (x+T,y+T), 255,4) x,y=1550,600 ; ROI[3,:,:] = image[y:y+T,x:x+T]; A_rgb_ROI = cv2.rectangle(A_rgb_ROI, (x,y), (x+T,y+T), 255,4) x,y=250,1400 ; ROI[4,:,:] = image[y:y+T,x:x+T]; A_rgb_ROI = cv2.rectangle(A_rgb_ROI, (x,y), (x+T,y+T), 255,4) x,y=250,600 ; ROI[5,:,:] = image[y:y+T,x:x+T]; A_rgb_ROI = cv2.rectangle(A_rgb_ROI, (x,y), (x+T,y+T), 255,4) x,y=0,800 ; ROI[6,:,:] = image[y:y+T,x:x+T]; A_rgb_ROI = cv2.rectangle(A_rgb_ROI, (x,y), (x+T,y+T), 0,4) plt.figure(1) plt.imshow(A_rgb_ROI);plt.axis('off') K = 7 Prop = np.empty((K,6)); for i in range(K) : ROI_image = ROI[i,:,:] glcm = feature.graycomatrix(ROI_image, distances=[1], angles=[0], levels=256, symmetric=True) prop {‘contrast’, ‘dissimilarity’, ‘homogeneity’, ‘energy’, ‘correlation’, ‘ASM’} Prop[i,0] = feature.graycoprops(glcm, 'contrast') Prop[i,1] = feature.graycoprops(glcm, 'dissimilarity') Prop[i,2] = feature.graycoprops(glcm, 'homogeneity') Prop[i,3] = feature.graycoprops(glcm, 'energy') Prop[i,4] = feature.graycoprops(glcm, 'correlation') Prop[i,5] = feature.graycoprops(glcm, 'ASM') Features plot plt.figure(2) Labels = ['contrast','dissimilarity','homogeneity','energy','correlation','ASM'] for i in range(K) : plt.subplot(2,7,i+1);plt.imshow(ROI[i,:,:], cmap='gray');plt.title('ROI %d'%i); plt.axis('off') for i in np.arange(6) : plt.subplot(2,6,7+i);plt.stem(Prop[:,i]);plt.title('%s'%Labels[i]) #%% K means Segmentation AV = A_rgb.reshape((-1,3)); AV = np.float32(AV) define criteria, number of clusters(K) and apply kmeans() criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 10, 1.0) K = 6 compac,label,center=cv2.kmeans(AV,K,None,criteria,10,cv2.KMEANS_RANDOM_CENTERS) center=np.round(center); center = np.uint8(center) AV_seg = center[label.flatten()] A_Seg = label.reshape((image.shape)) plt.figure(3) plt.subplot(321);plt.imshow(A_rgb);plt.title('Image originale');plt.axis('off') plt.subplot(322);plt.imshow(A_Seg, cmap='hsv');plt.axis('off') #%% GLCM Features extraction of ROI by masking Prop = np.empty((K,6)); for i in range(K) : #Create mask for region i mask = np.where(A_Seg == i, 1, 0); ROI_image = np.multiply(image.astype(np.double), mask).astype(np.uint8) plt.subplot(3,6,(i+7));plt.imshow(ROI_image, cmap='gray');plt.title('ROI');plt.axis('off') ROI_image = ROI[i,:,:] glcm = feature.graycomatrix(ROI_image, distances=[1], angles=[0], levels=256, symmetric=True) prop {‘contrast’, ‘dissimilarity’, ‘homogeneity’, ‘energy’, ‘correlation’, ‘ASM’} Prop[i,0] = feature.graycoprops(glcm, 'contrast') Prop[i,1] = feature.graycoprops(glcm, 'dissimilarity') Prop[i,2] = feature.graycoprops(glcm, 'homogeneity') Prop[i,3] = feature.graycoprops(glcm, 'energy') Prop[i,4] = feature.graycoprops(glcm, 'correlation') Prop[i,5] = feature.graycoprops(glcm, 'ASM') for i in np.arange(6) : plt.subplot(3,6,13+i);plt.stem(Prop[:,i]);plt.title('%s'%Labels[i]) #%% LPB Features radius = 1; n_points = 8 * radius ker = np.ones((5,5))/25 image = cv2.filter2D(src=image, ddepth=-1, kernel=ker); lbp8 = feature.local_binary_pattern(image, n_points,radius, method='default'); lbp8 =np.uint8 (lbp8) hist8, _ = np.histogram(lbp8.ravel(), bins=256, range=(0, 256)); lbp8RI = feature.local_binary_pattern(image, n_points,radius, method='ror'); lbp8 =np.uint8 (lbp8) hist8RI, _ = np.histogram(lbp8RI.ravel(), bins=256, range=(0, 256)); lbp8RIu = feature.local_binary_pattern(image, n_points,radius, method='uniform') hist8RIu, _ = np.histogram(lbp8RIu.ravel(),bins=np.arange(0, n_points + 3),range=(0, n_points + 2));hist8RIu = hist8RIu.astype("float"); hist8RIu /= hist8RIu.sum() plt.figure(4) plt.subplot(231);plt.imshow(image,cmap='gray');plt.title('Image'); plt.axis('off') plt.subplot(232);plt.imshow(lbp8,cmap='gray');plt.title('LBP'); plt.axis('off') plt.subplot(233);plt.imshow(lbp8RI,cmap='gray');plt.title('LBP Rotation Invariant'); plt.axis('off') plt.subplot(234);plt.imshow(lbp8RIu,cmap='gray');plt.title('LBP uniform'); plt.axis('off') plt.subplot(235);plt.stem(hist8); plt.title('LBP Histogram'); plt.subplot(236);plt.stem(hist8RIu); plt.title('LBP Rotation Invariant Uniform Histogram');