![\"\"](\"https:\/\/www.wsisp.com\/helps\/wp-content\/uploads\/2025\/08\/20250812154937-689b629184276.png\")
\u5c31\u662f\u62fc\u63a5\u5168\u666f\u56fe，\u662f\u56fe\u7247\u62fc\u63a5\u7684\u5e94\u7528\u4e4b\u4e00，\u624b\u673a\u62cd\u7167\u90fd\u6709\u5168\u666f\u62cd\u6444\u529f\u80fd<\/p>\n\u56fe\u7247\u62fc\u63a5（image stitching）\u5c31\u662f\u5c06\u7edf\u4e00\u573a\u666f\u7684\u4e0d\u540c\u62cd\u6444\u51fa\u7684\u56fe\u7247\u62fc\u63a5\u5230\u4e00\u8d77，\u5982\u56fe\u6240\u793a<\/p>\n
\u5c31\u662f\u62fc\u63a5\u5168\u666f\u56fe，\u662f\u56fe\u7247\u62fc\u63a5\u7684\u5e94\u7528\u4e4b\u4e00，\u624b\u673a\u62cd\u7167\u90fd\u6709\u5168\u666f\u62cd\u6444\u529f\u80fd<\/p>\n
\u4ed4\u7ec6\u89c2\u5bdf\u5168\u666f\u56fe，\u5bfb\u627e\u5b83\u4eec\u76f8\u4f3c\u6027，\u56fe8-2\u7684\u5168\u666f\u56fe\u53ef\u4ee5\u901a\u8fc7\u7f29\u653e，\u65cb\u8f6c，\u5c04\u5f71\u7b49\u64cd\u4f5c\u8fdb\u884c\u62fc\u63a5\u800c\u6210，\u6211\u4eec\u9996\u5148\u4ecb\u7ecd\u51e0\u4e2a\u5e38\u7528\u7684\u56fe\u50cf\u53d8\u6362<\/p>\n<\/p>\n
\u5e73\u79fb\u53d8\u6362\u901a\u8fc7\u5411\u91cf ( \\\\mathbf{t} = (t_x, t_y) ) \u5b9e\u73b0，\u56fe\u50cf\u4e0a\u70b9 ( \\\\mathbf{p} = (i, j) ) \u5e73\u79fb\u540e\u5f97\u5230\u65b0\u70b9 ( \\\\mathbf{p}' = (i', j') )，\u6ee1\u8db3： [ \\\\mathbf{p}' = \\\\mathbf{p} + \\\\mathbf{t} ] \u5176\u4e2d ( t_x ) \u548c ( t_y ) \u5206\u522b\u8868\u793a\u6c34\u5e73\u548c\u5782\u76f4\u65b9\u5411\u7684\u5e73\u79fb\u8ddd\u79bb\u3002<\/p>\n
\u65cb\u8f6c\u53d8\u6362\u7ed5\u539f\u70b9\u9006\u65f6\u9488\u65cb\u8f6c\u89d2\u5ea6 ( \\\\theta )，\u70b9 ( \\\\mathbf{p} = (i, j) ) \u65cb\u8f6c\u540e\u5f97\u5230 ( \\\\mathbf{p}' = R\\\\mathbf{p} )，\u65cb\u8f6c\u77e9\u9635 ( R ) \u4e3a： [ R = \\\\begin{bmatrix} \\\\cos \\\\theta & -\\\\sin \\\\theta \\\\ \\\\sin \\\\theta & \\\\cos \\\\theta \\\\end{bmatrix} ]<\/p>\n
\u4ee5\u539f\u70b9\u4e3a\u4e2d\u5fc3，\u6cbf ( x ) \u8f74\u7f29\u653e ( s_x ) \u500d，\u6cbf ( y ) \u8f74\u7f29\u653e ( s_y ) \u500d，\u70b9 ( \\\\mathbf{p} = (i, j) ) \u7f29\u653e\u540e\u5f97\u5230 ( \\\\mathbf{p}' = S\\\\mathbf{p} )，\u7f29\u653e\u77e9\u9635 ( S ) \u4e3a： [ S = \\\\begin{bmatrix} s_x & 0 \\\\ 0 & s_y \\\\end{bmatrix} ]<\/p>\n
\u5c04\u5f71\u53d8\u6362\u662f\u66f4\u4e00\u822c\u7684\u7ebf\u6027\u53d8\u6362，\u53ef\u7528\u9f50\u6b21\u5750\u6807\u8868\u793a\u3002\u5bf9\u4e8e\u70b9 ( \\\\mathbf{p} = (i, j, 1) )（\u9f50\u6b21\u5750\u6807），\u53d8\u6362\u540e ( \\\\mathbf{p}' = H\\\\mathbf{p} )，\u5176\u4e2d ( H ) \u4e3a ( 3 \\\\times 3 ) \u53d8\u6362\u77e9\u9635： [ H = \\\\begin{bmatrix} h_{11} & h_{12} & h_{13} \\\\ h_{21} & h_{22} & h_{23} \\\\ h_{31} & h_{32} & h_{33} \\\\end{bmatrix} ] \u5c04\u5f71\u53d8\u6362\u80fd\u5b9e\u73b0\u503e\u659c\u3001\u900f\u89c6\u7b49\u590d\u6742\u51e0\u4f55\u53d8\u6362\u3002<\/p>\n
\u56fe8-1\u7684\u5b50\u56fe\u4e0e\u56fe8-2\u5168\u666f\u56fe\u7684\u76f8\u4f3c\u6027\u4f53\u73b0\u5728：<\/p>\n
\u6240\u6709\u53d8\u6362\u5747\u53ef\u8868\u793a\u4e3a\u77e9\u9635\u4e58\u6cd5（\u9f50\u6b21\u5750\u6807\u4e0b）： [ \\\\mathbf{p}' = M\\\\mathbf{p} ] \u5176\u4e2d ( M ) \u4e3a\u5bf9\u5e94\u53d8\u6362\u77e9\u9635\u3002\u5e73\u79fb\u9700\u6269\u5c55\u4e3a\u4eff\u5c04\u53d8\u6362： [ M_{\\\\text{\u5e73\u79fb}} = \\\\begin{bmatrix} 1 & 0 & t_x \\\\ 0 & 1 & t_y \\\\ 0 & 0 & 1 \\\\end{bmatrix} ]<\/p>\n
1.\u901a\u8fc7SIFT\u8ba1\u7b97\u51fa\u4e24\u5e45\u56fe\u7247\u7684\u7279\u5f81\u70b9<\/p>\n
2.\u5c06\u4e24\u5e45\u56fe\u7247\u7684\u7279\u5f81\u70b9\u8fdb\u884c\u5339\u914d<\/p>\n
3.\u66f4\u5177\u5339\u914d\u7684\u7279\u5f81\u70b9\u8ba1\u7b97\u56fe\u7247\u53d8\u6362\u77e9\u9635<\/p>\n
![\"\"](\"https:\/\/www.wsisp.com\/helps\/wp-content\/uploads\/2025\/08\/20250812154940-689b6294da4d2.png\")
\n\u5f53\u5229\u7528SIFT\u8fdb\u884c\u7279\u5f81\u5339\u914d\u65f6，\u6709\u4e9b\u65f6\u5019\u53ef\u80fd\u4f1a\u51fa\u73b0\u56fe8-6\u7684\u60c5\u51b5\u3002\u56fe8-6\u4e2d\u53f3\u56fe\u7eff\u8272\u5706\u5708
\n\u5185\u7684\u7279\u5f81\u70b9\u662f\u4e0e\u5de6\u56fe\u5339\u914d\u7684\u7279\u5f81\u70b9，\u4f46\u5229\u7528SIFT\u5339\u914d\u7279\u5f81\u70b9\u65f6，\u4f1a\u5c06\u5de6\u56fe\u4e2d\u90e8\u5206\u7279\u5f81\u70b9\u5339\u914d\u5230
\n\u53f3\u56fe\u7eff\u8272\u5706\u5708\u4e4b\u5916\u7684\u7279\u5f81\u70b9（\u5982\u7ea2\u8272\u5706\u5708\u5185\u7684\u7279\u5f81\u70b9）\u3002\u8fd9\u4e9b\u7279\u5f81\u70b9\u5339\u914d\u662f\u9519\u8bef\u7684\u5339\u914d，\u5e94\u8be5\u88ab
\n\u79fb\u9664，\u4ece\u800c\u4fdd\u8bc1\u53d8\u6362\u77e9\u9635\u8ba1\u7b97\u7684\u9c81\u68d2\u6027\u3002\u5e94\u8be5\u5982\u4f55\u79fb\u9664\u9519\u8bef\u7684\u5339\u914d\u70b9\u5bf9\u5462？<\/p>\n
\u53ef\u4ee5\u7528\u5230RANSAC\u7b97\u6cd5<\/p>\n
RANSAC（Random Sample Consensus）\u662f\u4e00\u79cd\u9c81\u68d2\u7684\u6a21\u578b\u62df\u5408\u7b97\u6cd5，\u5e38\u7528\u4e8e\u5904\u7406\u5305\u542b\u5927\u91cf\u566a\u58f0\u6216\u5f02\u5e38\u503c\u7684\u6570\u636e\u3002\u5728\u8ba1\u7b97\u673a\u89c6\u89c9\u4e2d，RANSAC\u5e38\u7528\u4e8e\u53bb\u9664\u7279\u5f81\u5339\u914d\u4e2d\u7684\u8bef\u5339\u914d（outliers），\u4ec5\u4fdd\u7559\u6ee1\u8db3\u51e0\u4f55\u7ea6\u675f\u7684\u6b63\u786e\u5339\u914d（inliers）\u3002<\/p>\n
RANSAC\u901a\u8fc7\u968f\u673a\u91c7\u6837\u6700\u5c0f\u6570\u636e\u96c6\u8fed\u4ee3\u4f30\u8ba1\u6a21\u578b\u53c2\u6570，\u5e76\u7edf\u8ba1\u652f\u6301\u8be5\u6a21\u578b\u7684\u6837\u672c\u6570\u91cf\u3002\u7b97\u6cd5\u6838\u5fc3\u601d\u60f3\u662f：\u6b63\u786e\u7684\u5339\u914d\u5e94\u6ee1\u8db3\u67d0\u79cd\u51e0\u4f55\u53d8\u6362（\u5982\u5355\u5e94\u6027\u77e9\u9635\u6216\u57fa\u7840\u77e9\u9635），\u800c\u8bef\u5339\u914d\u5219\u4e0d\u7b26\u5408\u8be5\u7ea6\u675f\u3002<\/p>\n
\u8f93\u5165\u51c6\u5907<\/p>\n
\u6838\u5fc3\u6d41\u7a0b<\/p>\n
import cv2
\nimport numpy as np<\/p>\n
def ransac_filter_matches(points1, points2, model='homography', max_iter=1000, threshold=3.0):
\n """
\n points1, points2: \u5339\u914d\u7684\u70b9\u5750\u6807 (N\u00d72 numpy\u6570\u7ec4)
\n model: \u62df\u5408\u6a21\u578b\u7c7b\u578b ('homography' \u6216 'fundamental')
\n """
\n if len(points1) < 4:
\n return np.arange(len(points1)) # \u4e0d\u8db34\u5bf9\u70b9\u65f6\u8fd4\u56de\u6240\u6709\u7d22\u5f15<\/p>\n
if model == 'homography':
\n H, mask = cv2.findHomography(points1, points2, cv2.RANSAC, threshold, maxIters=max_iter)
\n elif model == 'fundamental':
\n F, mask = cv2.findFundamentalMat(points1, points2, cv2.FM_RANSAC, threshold, max_iter)<\/p>\n
return mask.ravel().astype(bool) # \u8fd4\u56de\u5185\u70b9\u63a9\u7801<\/p>\n
\u9608\u503c\u9009\u62e9：\u901a\u5e38\u8bbe\u7f6e\u4e3a1-5\u50cf\u7d20，\u53d6\u51b3\u4e8e\u7279\u5f81\u70b9\u5b9a\u4f4d\u7cbe\u5ea6\u3002\u5bf9\u4e8eSIFT\/SURF\u7b49\u7279\u5f81\u53ef\u8bbe\u4e3a3，ORB\u7b49\u4e8c\u8fdb\u5236\u7279\u5f81\u5efa\u8bae\u8bbe\u4e3a5<\/p>\n<\/li>\n
\u8fed\u4ee3\u6b21\u6570：\u9ed8\u8ba41000\u6b21\u53ef\u6ee1\u8db3\u5927\u591a\u6570\u573a\u666f\u3002\u53ef\u901a\u8fc7\u516c\u5f0f\u4f30\u7b97：<\/p>\n
$$ N = \\\\frac{\\\\log(1-p)}{\\\\log(1-(1-\\\\epsilon)^s)} $$<\/p>\n
\u5176\u4e2dp\u4e3a\u7f6e\u4fe1\u5ea6（\u59820.99），\u03b5\u4e3a\u5f02\u5e38\u503c\u6bd4\u4f8b\u4f30\u8ba1\u503c，s\u4e3a\u6700\u5c0f\u6837\u672c\u6570<\/p>\n<\/li>\n<\/ul>\n
# \u5047\u8bbe\u5df2\u6709\u5339\u914d\u7ed3\u679c
\nmatches = flann.knnMatch(des1, des2, k=2)
\ngood_matches = [m for m,n in matches if m.distance < 0.7*n.distance]<\/p>\n
# \u63d0\u53d6\u5339\u914d\u70b9\u5750\u6807
\npts1 = np.float32([kp1[m.queryIdx].pt for m in good_matches]).reshape(-1,2)
\npts2 = np.float32([kp2[m.trainIdx].pt for m in good_matches]).reshape(-1,2)<\/p>\n
# RANSAC\u8fc7\u6ee4
\ninlier_mask = ransac_filter_matches(pts1, pts2)
\nfinal_matches = [good_matches[i] for i in range(len(good_matches)) if inlier_mask[i]]<\/p>\n
\u56fe\u50cf\u62fc\u63a5\u7684\u6700\u540e\u4e00\u6b65\u662f\u5c06\u8f93\u5165\u56fe\u50cf\u53d8\u6362\u5e76\u7f1d\u5408\u5230\u4e00\u5e45\u56fe\u50cf\u4e2d\u3002\u5bf9\u4e8e\u4e24\u5e45\u56fe\u50cfA\u548cB，\u5728\u5df2
\n\u7ecf\u68c0\u6d4b\u51fa\u5bf9\u5e94\u7684\u7279\u5f81\u70b9\u5bf9，\u5e76\u5229\u7528RANSAC\u7b97\u6cd5\u8ba1\u7b97\u5f97\u5230\u53d8\u6362\u77e9\u9635T\u4e4b\u540e，\u5c06\u56fe\u50cfB\u8f6c\u6362\u4e3a
\nTB\u3002\u7136\u540e，\u5bf9\u8f6c\u6362\u540e\u7684\u56fe\u50cf，\u5373TB，\u4e0e\u56fe\u50cfA\u5728\u91cd\u53e0\u90e8\u5206\u7684\u50cf\u7d20\u503c\u6c42\u5e73\u5747\u503c，\u4ee5\u4f18\u5316\u56fe\u50cf\u7f1d
\n\u5408\u7684\u8fb9\u754c\u3002\u5982\u6b64，\u4fbf\u53ef\u5f97\u5230\u6700\u7ec8\u7f1d\u5408\u597d\u7684\u62fc\u63a5\u56fe\u50cf\u3002
\n\u7efc\u4e0a\u6240\u8ff0，\u6211\u4eec\u628a\u56fe\u50cf\u62fc\u63a5\u7684\u5168\u8fc7\u7a0b\u603b\u7ed3\u4e3a\u4ee5\u4e0b4\u6b65：
\n（1）\u8ba1\u7b97\u4e24\u5e45\u56fe\u50cf\u7684\u7279\u5f81\u70b9；
\n（2）\u5c06\u4e24\u5e45\u56fe\u50cf\u7684\u7279\u5f81\u70b9\u8fdb\u884c\u5339\u914d；
\n（3）\u6839\u636e\u5339\u914d\u7684\u7279\u5f81\u70b9\u5bf9，\u5229\u7528RANSAC\u7b97\u6cd5\u8ba1\u7b97\u56fe\u50cf\u53d8\u6362\u77e9\u9635；
\n（4）\u5c06\u56fe\u50cf\u8fdb\u884c\u62fc\u63a5\u3002<\/p>\n
import cv2<\/p>\n
# \u8bfb\u53d6\u56fe\u50cf
\nimage1 = cv2.imread('image1.jpeg')
\nimage2 = cv2.imread('image2.jpeg')<\/p>\n
# \u68c0\u67e5\u56fe\u50cf\u662f\u5426\u6210\u529f\u8bfb\u53d6
\nif image1 is None or image2 is None:
\n print("\u65e0\u6cd5\u8bfb\u53d6\u56fe\u50cf\u6587\u4ef6")
\n exit()<\/p>\n
# \u521b\u5efa\u62fc\u63a5\u5668 \u6548\u679c：\u62fc\u63a5\u7ed3\u679c\u51fa\u73b0\u4e86\u8fb9\u7f18\u9ed1\u8fb9\u548c\u5f62\u53d8
\nstitcher = cv2.Stitcher_create() if hasattr(cv2, 'Stitcher_create') else cv2.createStitcher()<\/p>\n
# \u6267\u884c\u62fc\u63a5
\n(status, stitched) = stitcher.stitch([image1, image2])<\/p>\n
# \u4fdd\u5b58\u7ed3\u679c
\nif status == cv2.Stitcher_OK:
\n cv2.imwrite('stitched_output.jpg', stitched)
\n print("\u62fc\u63a5\u6210\u529f，\u7ed3\u679c\u5df2\u4fdd\u5b58\u4e3a 'stitched_output.jpg'")
\nelse:
\n print(f'\u62fc\u63a5\u5931\u8d25，\u9519\u8bef\u4ee3\u7801: {status}')<\/p>\n
import cv2
\nimport numpy as np<\/p>\n
def stitch_images(images, ratio=0.75, reproj_thresh=4.0, show_matches=False):
\n """
\n \u56fe\u50cf\u62fc\u63a5\u51fd\u6570<\/p>\n
\u53c2\u6570:
\n images: \u8981\u62fc\u63a5\u7684\u56fe\u50cf\u5217\u8868
\n ratio: Lowe's ratio test\u53c2\u6570
\n reproj_thresh: RANSAC\u91cd\u6295\u5f71\u9608\u503c
\n show_matches: \u662f\u5426\u663e\u793a\u7279\u5f81\u5339\u914d\u7ed3\u679c<\/p>\n
\u8fd4\u56de:
\n \u62fc\u63a5\u540e\u7684\u56fe\u50cf
\n """
\n # \u521d\u59cb\u5316OpenCV\u7684SIFT\u7279\u5f81\u68c0\u6d4b\u5668
\n sift = cv2.SIFT_create()<\/p>\n
# \u68c0\u6d4b\u5173\u952e\u70b9\u548c\u63cf\u8ff0\u7b26
\n (kpsA, featuresA) = sift.detectAndCompute(images[0], None)
\n (kpsB, featuresB) = sift.detectAndCompute(images[1], None)<\/p>\n
# \u5339\u914d\u7279\u5f81\u70b9
\n matcher = cv2.DescriptorMatcher_create("BruteForce")
\n raw_matches = matcher.knnMatch(featuresA, featuresB, 2)<\/p>\n
# \u5e94\u7528Lowe's ratio test\u7b5b\u9009\u597d\u7684\u5339\u914d\u70b9
\n good_matches = []
\n for m in raw_matches:
\n if len(m) == 2 and m[0].distance < m[1].distance * ratio:
\n good_matches.append((m[0].trainIdx, m[0].queryIdx))<\/p>\n
# \u81f3\u5c11\u9700\u89814\u4e2a\u5339\u914d\u70b9\u624d\u80fd\u8ba1\u7b97\u5355\u5e94\u6027\u77e9\u9635
\n if len(good_matches) > 4:
\n ptsA = np.float32([kpsA[i].pt for (_, i) in good_matches])
\n ptsB = np.float32([kpsB[i].pt for (i, _) in good_matches])<\/p>\n
# \u8ba1\u7b97\u5355\u5e94\u6027\u77e9\u9635
\n (H, status) = cv2.findHomography(ptsA, ptsB, cv2.RANSAC, reproj_thresh)<\/p>\n
# \u62fc\u63a5\u56fe\u50cf
\n result = cv2.warpPerspective(images[0], H,
\n (images[0].shape[1] + images[1].shape[1],
\n images[0].shape[0]))
\n result[0:images[1].shape[0], 0:images[1].shape[1]] = images[1]<\/p>\n
# \u5982\u679c\u9700\u8981\u663e\u793a\u5339\u914d\u7ed3\u679c
\n if show_matches:
\n vis = np.zeros((max(images[0].shape[0], images[1].shape[0]),
\n images[0].shape[1] + images[1].shape[1], 3), dtype=np.uint8)
\n vis[0:images[0].shape[0], 0:images[0].shape[1]] = images[0]
\n vis[0:images[1].shape[0], images[0].shape[1]:] = images[1]<\/p>\n
for ((trainIdx, queryIdx), s) in zip(good_matches, status):
\n if s == 1:
\n ptA = (int(kpsA[queryIdx].pt[0]), int(kpsA[queryIdx].pt[1]))
\n ptB = (int(kpsB[trainIdx].pt[0]) + images[0].shape[1],
\n int(kpsB[trainIdx].pt[1]))
\n cv2.line(vis, ptA, ptB, (0, 255, 0), 1)<\/p>\n
cv2.imshow("Feature Matches", vis)
\n cv2.waitKey(0)
\n cv2.destroyAllWindows()<\/p>\n
return result<\/p>\n
return None<\/p>\n
# \u793a\u4f8b\u7528\u6cd5
\nif __name__ == "__main__":
\n # \u8bfb\u53d6\u4e24\u5f20\u8981\u62fc\u63a5\u7684\u56fe\u50cf
\n image1 = cv2.imread("image1.jpeg")
\n image2 = cv2.imread("image2.jpeg")<\/p>\n
# \u786e\u4fdd\u56fe\u50cf\u8bfb\u53d6\u6210\u529f
\n if image1 is None or image2 is None:
\n print("\u65e0\u6cd5\u8bfb\u53d6\u56fe\u50cf\u6587\u4ef6")
\n exit()<\/p>\n
# \u8c03\u6574\u56fe\u50cf\u5927\u5c0f(\u53ef\u9009)
\n image1 = cv2.resize(image1, (0, 0), fx=0.5, fy=0.5)
\n image2 = cv2.resize(image2, (0, 0), fx=0.5, fy=0.5)<\/p>\n
# \u62fc\u63a5\u56fe\u50cf
\n stitched_image = stitch_images([image1, image2], show_matches=True)<\/p>\n
if stitched_image is not None:
\n # \u663e\u793a\u5e76\u4fdd\u5b58\u7ed3\u679c
\n cv2.imshow("Stitched Image", stitched_image)
\n cv2.waitKey(0)
\n cv2.destroyAllWindows()
\n cv2.imwrite("stitched_result.jpg", stitched_image)
\n else:
\n print("\u56fe\u50cf\u62fc\u63a5\u5931\u8d25，\u53ef\u80fd\u5339\u914d\u70b9\u4e0d\u8db3")<\/p>\n
\u4e24\u79cd\u65b9\u6cd5\u90fd\u9700\u8981\u5b89\u88c5OpenCV：<\/p>\n
pip install opencv-python opencv-contrib-python<\/p>\n","protected":false},"excerpt":{"rendered":"
\u6587\u7ae0\u6d4f\u89c8\u9605\u8bfb290\u6b21\uff0c\u70b9\u8d5e2\u6b21\uff0c\u6536\u85cf2\u6b21\u3002\u56fe\u7247\u62fc\u63a5\uff08image stitching\uff09\u5c31\u662f\u5c06\u7edf\u4e00\u573a\u666f\u7684\u4e0d\u540c\u62cd\u6444\u51fa\u7684\u56fe\u7247\u62fc\u63a5\u5230\u4e00\u8d77\uff0c\u5982\u56fe\u6240\u793a\u5c31\u662f\u62fc\u63a5\u5168\u666f\u56fe\uff0c\u662f\u56fe\u7247\u62fc\u63a5\u7684\u5e94\u7528\u4e4b\u4e00\uff0c\u624b\u673a\u62cd\u7167\u90fd\u6709\u5168\u666f\u62cd\u6444\u529f\u80fd\u4ed4\u7ec6\u89c2\u5bdf\u5168\u666f\u56fe\uff0c\u5bfb\u627e\u5b83\u4eec\u76f8\u4f3c\u6027\uff0c\u56fe8-2\u7684\u5168\u666f\u56fe\u53ef\u4ee5\u901a\u8fc7\u7f29\u653e\uff0c\u65cb\u8f6c\uff0c\u5c04\u5f71\u7b49\u64cd\u4f5c\u8fdb\u884c\u62fc\u63a5\u800c\u6210\uff0c\u6211\u4eec\u9996\u5148\u4ecb\u7ecd\u51e0\u4e2a\u5e38\u7528\u7684\u56fe\u50cf\u53d8\u6362RANSAC\uff08Random Sample Consensus\uff09\u662f\u4e00\u79cd\u9c81\u68d2\u7684\u6a21\u578b\u62df\u5408\u7b97\u6cd5\uff0c\u5e38\u7528\u4e8e\u5904\u7406\u5305\u542b\u5927\u91cf\u566a\u58f0\u6216\u5f02\u5e38\u503c\u7684\u6570\u636e\u3002<\/p>\n","protected":false},"author":2,"featured_media":54198,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[50,523],"topic":[],"class_list":["post-54200","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-server","tag-50","tag-523"],"yoast_head":"\n