opencv完成多幅图像的拼接

原理

• 角点特征匹配
• 单应性变换Homography：$\begin{bmatrix} x_i^\prime \\ y_i^\prime \\1\end{bmatrix}\cong \begin{bmatrix} h_{00} & h_{01} & h_{02} \\h_{10} & h_{11} & h_{12} \\h_{20} & h_{21} & h_{22}\end{bmatrix}\begin{bmatrix} x_i \\ y_i \\1\end{bmatrix}$。这里是近似等于，实际上他们之间有比例关系$\lambda p=HQ$

变换	矩阵	自由度数量
Translation		2
Similarity	$\begin{bmatrix} sR & t \\ 0^T & 1 \end{bmatrix}$	4
Afine	$\begin{bmatrix} A & t \\ 0^T & 1 \end{bmatrix}$	6
Homography		8，$h_{22}$为1
Euclidean	$\begin{bmatrix} R & t \\ 0^T & 1 \end{bmatrix}$	3
Projective	$\begin{bmatrix} A & t \\ v^T & 1 \end{bmatrix}$

实现1

数据

代码

#include <iostream>
#include <opencv2/opencv.hpp>
using namespace std;
using namespace cv;
int main()
{
	vector< string> files;
	glob("./image\\", files);
	vector<Mat> images;
	for (size_t i = 0; i < files.size(); i++) {
		printf("image file : %s n", files[i].c_str());
		images.push_back(imread(files[i]));
	}
	for (size_t t = 0; t < images.size(); t++) {
		namedWindow("input" + to_string(t));
		imshow("input" + to_string(t), images[t]);
	}
	// 设置拼接模式与参数
	Mat result1, result2, result3;
	Stitcher::Mode mode = Stitcher::PANORAMA;
	Ptr<Stitcher> stitcher = Stitcher::create(mode);
	// 拼接方式-多通道融合
	auto blender = detail::Blender::createDefault(detail::Blender::MULTI_BAND);
	stitcher->setBlender(blender);
	// 拼接
	Stitcher::Status status = stitcher->stitch(images, result1);
	// 平面曲翘拼接
	//auto plane_warper = makePtr<cv::PlaneWarper>;
	//stitcher->setWarper(plane_warper);
	//status = stitcher->stitch(images, result2);
	// 鱼眼拼接
	//auto fisheye_warper = makePtr<cv::FisheyeWarper>;
	//stitcher->setWarper(fisheye_warper);
	//status = stitcher->stitch(images, result3);
	// 检查返回
	if (status != Stitcher::OK)
	{
		cout << "Can't stitch images, error code = " << int(status) << endl;
		return EXIT_FAILURE;
	}
	imwrite("output01.png", result1);
	//imwrite("output02.png", result2);
	//imwrite("output03.png", result3);
	waitKey(0);
	return 0;
}

输出

实现2

数据

代码

import matplotlib.pyplot as plt
import numpy as np
import cv2
img1 = cv2.imread(r"images/contract_left.png")
img1 = cv2.cvtColor(img1,cv2.COLOR_BGR2RGB)
img2 = cv2.imread(r"images/contract_right.png")
img2 = cv2.cvtColor(img2,cv2.COLOR_BGR2RGB)
fig = plt.figure(figsize=(18,18))
a = fig.add_subplot(1,2,1)
imgplot = plt.imshow(img1)
a.set_title("11")
a = fig.add_subplot(1,2,2)
imgplot = plt.imshow(img2)
a.set_title("12")
def detectAndDesc(img):
    img_gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
    descripter = cv2.SIFT_create()
    kps,desc = descripter.detectAndCompute(img,None)
    kps = np.float32([kp.pt for kp in kps])
    return kps,desc
def matchKeyPoints(kp1,des1,kp2,des2,ratio=0.75,reprojThresh=4.0):
    matcher = cv2.BFMatcher()
    rawMatches = matcher.knnMatch(des1,des2,k=2)
    matches = []
    for n in rawMatches:
        if len(n)==2 and n[0].distance< n[1].distance*ratio:
            matches.append((n[0].trainIdx,n[0].queryIdx))
    if len(matches)>4:
        p1 = np.float32([kp1[i] for (_,i) in matches])
        p2 = np.float32([kp2[i] for (i,_) in matches])
        H,status = cv2.findHomography(p1,p2,cv2.RANSAC,reprojThresh)
        return matches,H,status
    else:
        return None,None,False
kp1,des1 = detectAndDesc(img1)
kp2,des2 = detectAndDesc(img2)
ratio = 0.75
reprojThresh = 4.0
matches,H,status = matchKeyPoints(kp1,des1,kp2,des2,ratio,reprojThresh)
result = cv2.warpPerspective(img1,H,(img1.shape[1]+img2.shape[1],img1.shape[0]))
fig = plt.figure(figsize=(18,18))
a = fig.add_subplot(2,1,1)
imgplot = plt.imshow(result)
a.set_title("11")
result[0:img2.shape[0],0:img2.shape[1]]=img2
a = fig.add_subplot(2,1,2)
imgplot = plt.imshow(result)
a.set_title("12")

输出