######################################################################## # # Taking the1owls Image Matching script and experimenting; # - downsizing images to speed up # # Want to extract the warp parameters # ######################################################################## import numpy as np import cv2 import matplotlib.pyplot as plt print(cv2.__version__) # def im_stitcher(imp1, imp2, pcntDownsize = 1.0, withTransparency=False): # #Read image1 # image1 = cv2.imread(imp1) # # perform the resizing of the image by pcntDownsize and create a Grayscale version # dim1 = (int(image1.shape[1] * pcntDownsize), int(image1.shape[0] * pcntDownsize)) # img1 = cv2.resize(image1, dim1, interpolation = cv2.INTER_AREA) # img1Gray = cv2.cvtColor(img1, cv2.COLOR_BGR2GRAY) # #Read image2 # image2 = cv2.imread(imp2) # # perform the resizing of the image by pcntDownsize and create a Grayscale version # dim2 = (int(image2.shape[1] * pcntDownsize), int(image2.shape[0] * pcntDownsize)) # img2 = cv2.resize(image2, dim2, interpolation = cv2.INTER_AREA) # img2Gray = cv2.cvtColor(img2, cv2.COLOR_BGR2GRAY) # #use BRISK to create keypoints in each image # brisk = cv2.BRISK_create() # kp1, des1 = brisk.detectAndCompute(img1Gray,None) # kp2, des2 = brisk.detectAndCompute(img2Gray,None) # # use BruteForce algorithm to detect matches among image keypoints # dm = cv2.DescriptorMatcher_create("BruteForce-Hamming") # matches = dm.knnMatch(des1,des2, 2) # matches_ = [] # for m in matches: # if len(m) == 2 and m[0].distance < m[1].distance * 0.75: # matches_.append((m[0].trainIdx, m[0].queryIdx)) # kp1_ = np.float32([kp1[m[1]].pt for m in matches_]).reshape(-1,1,2) # kp2_ = np.float32([kp2[m[0]].pt for m in matches_]).reshape(-1,1,2) # H, mask = cv2.findHomography(kp2_,kp1_, cv2.RANSAC, 4.0) # h1,w1 = img1.shape[:2] # h2,w2 = img2.shape[:2] # pts1 = np.float32([[0,0],[0,h1],[w1,h1],[w1,0]]).reshape(-1,1,2) # pts2 = np.float32([[0,0],[0,h2],[w2,h2],[w2,0]]).reshape(-1,1,2) # pts2_ = cv2.perspectiveTransform(pts2, H) # pts = np.concatenate((pts1, pts2_), axis=0) # [xmin, ymin] = np.int32(pts.min(axis=0).ravel() - 0.5) # [xmax, ymax] = np.int32(pts.max(axis=0).ravel() + 0.5) # t = [-xmin,-ymin] # Ht = np.array([[1,0,t[0]],[0,1,t[1]],[0,0,1]]) # #warp the colour version of image2 # im = cv2.warpPerspective(img2, Ht.dot(H), (xmax-xmin, ymax-ymin)) # #overlay colur version of image1 to warped image2 # if withTransparency == True: # h3,w3 = im.shape[:2] # bim = np.zeros((h3,w3,3), np.uint8) # bim[t[1]:h1+t[1],t[0]:w1+t[0]] = img1 # #imGray = cv2.cvtColor(im, cv2.COLOR_BGR2GRAY) # #imColor = cv2.applyColorMap(imGray, cv2.COLORMAP_JET) # #im =(im[:,:,2] - bim[:,:,2]) # im = cv2.addWeighted(im,0.6,bim,0.6,0) # else: # im[t[1]:h1+t[1],t[0]:w1+t[0]] = img1 # return(im) # ########################################################## # # # # Match all combinations of one set of images # # # ########################################################## # #img104_1_166_1 = im_stitcher("../input/test_sm/set104_1.jpeg", "../input/test_sm/set166_1.jpeg", 0.4, True) # #img104_2_166_4 = im_stitcher("../input/test_sm/set104_2.jpeg", "../input/test_sm/set166_4.jpeg", 0.4, True) # #img104_3_166_5 = im_stitcher("../input/test_sm/set104_3.jpeg", "../input/test_sm/set166_5.jpeg", 0.4, True) # #img104_4_166_3 = im_stitcher("../input/test_sm/set104_4.jpeg", "../input/test_sm/set166_3.jpeg", 0.4, True) # #img104_5_166_2 = im_stitcher("../input/test_sm/set104_5.jpeg", "../input/test_sm/set166_2.jpeg", 0.4, True) # #plt.imsave('Set104_1_166_1_BRISK_matching.jpeg',img104_1_166_1) # #plt.imsave('Set104_2_166_4_BRISK_matching.jpeg',img104_2_166_4) # #plt.imsave('Set104_3_166_5_BRISK_matching.jpeg',img104_3_166_5) # #plt.imsave('Set104_4_166_3_BRISK_matching.jpeg',img104_4_166_3) # #plt.imsave('Set104_6_166_2_BRISK_matching.jpeg',img104_5_166_2) # #img1_1_85_5 = im_stitcher("../input/test_sm/set1_1.jpeg", "../input/test_sm/set85_5.jpeg", 0.4, True) # #img1_2_85_4 = im_stitcher("../input/test_sm/set1_2.jpeg", "../input/test_sm/set85_4.jpeg", 0.4, True) # #img1_3_85_2 = im_stitcher("../input/test_sm/set1_3.jpeg", "../input/test_sm/set85_2.jpeg", 0.4, True) # #img1_4_85_3 = im_stitcher("../input/test_sm/set1_4.jpeg", "../input/test_sm/set85_3.jpeg", 0.4, True) # #img1_5_85_1 = im_stitcher("../input/test_sm/set1_5.jpeg", "../input/test_sm/set85_1.jpeg", 0.4, True) # #plt.imsave('Set1_1_85_5_BRISK_matching.jpeg',img1_1_85_5) # #plt.imsave('Set1_2_85_4_BRISK_matching.jpeg',img1_2_85_4) # #plt.imsave('Set1_3_85_2_BRISK_matching.jpeg',img1_3_85_2) # #plt.imsave('Set1_4_85_3_BRISK_matching.jpeg',img1_4_85_3) # #plt.imsave('Set1_5_85_1_BRISK_matching.jpeg',img1_5_85_1) # #img3_1_22_1 = im_stitcher("../input/test_sm/set3_1.jpeg", "../input/test_sm/set22_1.jpeg", 0.4, True) # #img3_2_22_2 = im_stitcher("../input/test_sm/set3_2.jpeg", "../input/test_sm/set22_2.jpeg", 0.4, True) # #img3_3_22_5 = im_stitcher("../input/test_sm/set3_3.jpeg", "../input/test_sm/set22_5.jpeg", 0.4, True) # #img3_4_22_3 = im_stitcher("../input/test_sm/set3_4.jpeg", "../input/test_sm/set22_3.jpeg", 0.4, True) # #img3_5_22_4 = im_stitcher("../input/test_sm/set3_5.jpeg", "../input/test_sm/set22_4.jpeg", 0.4, True) # #plt.imsave('Set3_1_22_1_BRISK_matching.jpeg',img3_1_22_1) # #plt.imsave('Set3_2_22_2_BRISK_matching.jpeg',img3_2_22_2) # #plt.imsave('Set3_3_22_5_BRISK_matching.jpeg',img3_3_22_5) # #plt.imsave('Set3_4_22_3_BRISK_matching.jpeg',img3_4_22_3) # #plt.imsave('Set3_5_22_4_BRISK_matching.jpeg',img3_5_22_4) # #img5_1_68_3 = im_stitcher("../input/train_sm/set5_1.jpeg", "../input/test_sm/set68_3.jpeg", 0.4, True) # #plt.imsave('Set5_1_68_3_BRISK_matching.jpeg',img5_1_68_3) # img160_5_74_1 = im_stitcher("../input/train_sm/set160_5.jpeg", "../input/test_sm/set74_1.jpeg", 1, True) # img160_5_74_2 = im_stitcher("../input/train_sm/set160_5.jpeg", "../input/test_sm/set74_2.jpeg", 0.4, True) # img160_5_74_3 = im_stitcher("../input/train_sm/set160_5.jpeg", "../input/test_sm/set74_3.jpeg", 0.4, True) # img160_5_74_4 = im_stitcher("../input/train_sm/set160_5.jpeg", "../input/test_sm/set74_4.jpeg", 0.4, True) # img160_5_74_5 = im_stitcher("../input/train_sm/set160_5.jpeg", "../input/test_sm/set74_5.jpeg", 0.4, True) # plt.imsave('Set160_5_74_1_BRISK_matching.jpeg',img160_5_74_1) # plt.imsave('Set160_5_74_2_BRISK_matching.jpeg',img160_5_74_2) # plt.imsave('Set160_5_74_3_BRISK_matching.jpeg',img160_5_74_3) # plt.imsave('Set160_5_74_4_BRISK_matching.jpeg',img160_5_74_4) # plt.imsave('Set160_5_74_5_BRISK_matching.jpeg',img160_5_74_5) plt.imsave("160_5",cv2.imread("../input/train_sm/set160_5.jpeg")) plt.imsave("74_1",cv2.imread("../input/test_sm/set74_1.jpeg"))