基于Harris角点的多视角图像全景拼接算法matlab仿真

本文主要是介绍基于Harris角点的多视角图像全景拼接算法matlab仿真，希望对大家解决编程问题提供一定的参考价值，需要的开发者们随着小编来一起学习吧！

1.算法运行效果图预览

2.算法运行软件版本

3.部分核心程序

4.算法理论概述

4.1 Harris角点检测

4.2 图像配准

4.3 图像变换和拼接

4.4 全景图像优化

5.算法完整程序工程

1.算法运行效果图预览

2.算法运行软件版本

matlab2022a

3.部分核心程序

 
function [ImageB] = func_stitch(ImageA, ImageB)% 获取ImageA和ImageB的尺寸  RR1 = size(ImageA,1);CC1 = size(ImageA,2);RR2 = size(ImageB,1);CC2 = size(ImageB,2);%角点检测  jiaodian_imageB    = corner_detector(ImageB);%提取角点  [xh,yh,~]          = anms(jiaodian_imageB, 300);%对提取的角点计算特征描述子[Pcorner]          = feat_desc_geoblur(ImageB, xh, yh);%角点检测 jiaodian_imageA    = corner_detector(ImageA);%提取角点[xh2, yh2, ~]      = anms(jiaodian_imageA, 300);%对提取的角点计算特征描述子[Pcorner2]         = feat_desc_geoblur(ImageA, xh2, yh2);% 对两张图像的特征描述子进行匹配[Pmatch]           = feat_match(Pcorner2, Pcorner);% 提取匹配成功的点  X_mh1 = xh2(Pmatch ~= -1);Y_mh1 = yh2(Pmatch ~= -1);X_mh2 = xh(Pmatch(Pmatch ~= -1));Y_mh2 = yh(Pmatch(Pmatch ~= -1));% 剔除误匹配点 [H, ~] = ransac_est_homography(X_mh1, Y_mh1, X_mh2, Y_mh2, 10);% 计算两张图像到各自边界的距离  ImageA_dist = dist2border(ImageA);ImageB_dist = dist2border(ImageB);% 计算img_i四个角的映射坐标 ul = H*[1 1 1]'; ul = ul/ul(end);ur = H*[CC1, 1, 1]'; ur = ur/ur(end);bl = H*[1, RR1, 1]'; bl = bl/bl(end);br = H*[CC1, RR1, 1]'; br = br/br(end);% 根据映射坐标确定需要填充的边界  Edge_up    = 0;  Edge_L     = 0;Edge_dw    = 0;Edge_R     = 0;if max(br(1),ur(1)) > CC2Edge_R = round(max(br(1),ur(1))-CC2+30);ImageB = padarray(ImageB, [0, Edge_R], 'post');endif max(br(2), bl(2)) > RR2Edge_dw = round(max(br(2), bl(2))-RR2+30);ImageB = padarray(ImageB, [Edge_dw, 0], 'post');endif min(ul(1), bl(1)) <= 0 Edge_L = round(-min(ul(1), bl(1))+30);ImageB = padarray(ImageB, [0, Edge_L], 'pre');endif min(ul(2), ur(2)) <= 0Edge_up = round(-min(ul(2), ur(2)) + 30);ImageB = padarray(ImageB, [Edge_up, 0], 'pre');end% 计算单应性矩阵的逆H_inv = inv(H);% 创建一个网格，用于映射img_b上的每个像素到img_i上的坐标    [yh, xh] = meshgrid(round(Edge_up+min(ul(2), ur(2))):round(Edge_up+max(bl(2), br(2))), round(Edge_L+min(ul(1), bl(1))):round(Edge_L+max(br(1),ur(1))));yh = yh(:); xh = xh(:);xy  = H_inv*[xh - Edge_L, yh - Edge_up, ones(size(xh,1),1)]';xh2 = int64(xy(1,:)'./xy(3,:)'); yh2 = int64(xy(2,:)'./xy(3,:)');
................................................................................
end
0102