### 完整Python代码实现OpenCV图像拼接 以下是使用OpenCV库完成图像拼接的完整代码，其中包括SIFT角点检测、Brute-Force匹配以及图像融合功能。 ```python import numpy as np import cv2 from matplotlib import pyplot as plt def detect_and_describe(image): """ 检测关键点并提取局部不变描述符 """ gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) sift = cv2.SIFT_create() (keypoints, descriptors) = sift.detectAndCompute(gray, None) return keypoints, descriptors def match_keypoints(keypointsA, keypointsB, descriptorsA, descriptorsB): """ 使用Brute-Force匹配特征点 """ bf = cv2.BFMatcher(cv2.NORM_L2, crossCheck=False) raw_matches = bf.knnMatch(descriptorsA, descriptorsB, k=2) # 应用Lowe's ratio test筛选高质量匹配 good = [] for m, n in raw_matches: if m.distance < 0.7 * n.distance: good.append((m.trainIdx, m.queryIdx)) return good def compute_homography(pointsA, pointsB): """ 计算单应矩阵 """ (H, status) = cv2.findHomography(pointsA, pointsB, cv2.RANSAC, reprojThresh=4.0) return H, status def stitch_images(imageA, imageB, H): """ 图像融合 """ result = cv2.warpPerspective(imageA, H, (imageA.shape[1] + imageB.shape[1], max(imageA.shape[0], imageB.shape[0]))) result[0:imageB.shape[0], 0:imageB.shape[1]] = imageB return result # 加载两张图片 imageA = cv2.imread('left_image.jpg') imageB = cv2.imread('right_image.jpg') # 提取关键点和描述符 keypointsA, descriptorsA = detect_and_describe(imageA)[^4] keypointsB, descriptorsB = detect_and_describe(imageB)[^4] # 特征匹配 matches = match_keypoints(keypointsA, keypointsB, descriptorsA, descriptorsB)[^2] if len(matches) > 4: ptsA = np.float32([keypointsA[i].pt for (_, i) in matches]) ptsB = np.float32([keypointsB[i].pt for (i, _) in matches]) # 计算单应矩阵 H, _ = compute_homography(ptsA, ptsB)[^4] # 图像融合 stitched_image = stitch_images(imageA, imageB, H)[^4] # 显示结果 plt.imshow(cv2.cvtColor(stitched_image, cv2.COLOR_BGR2RGB)) plt.axis('off') plt.show() else: print("Not enough matches are found.") ``` --- #### 关键技术说明： 1. **SIFT角点检测** 使用`cv2.SIFT_create()`创建SIFT检测器，并通过`detectAndCompute`方法提取关键点及其描述符[^4]。 2. **Brute-Force匹配** 利用`cv2.BFMatcher`进行暴力匹配，并采用Lowe’s Ratio Test剔除错误匹配[^2]。 3. **单应矩阵计算** 基于匹配的关键点对，利用`cv2.findHomography`函数并通过RANSAC算法估算单应矩阵[^4]。 4. **图像融合** 使用`cv2.warpPerspective`将一张图像投影到另一张图像的空间坐标系下，最终叠加形成全景图。 --- ###