MATLAB特征点的检测与提取（2）

2，基于点特征匹配的杂乱场景目标检测

第一步：读入图片；

wahaha1 = imread('wahaha.jpg');
wahaha=rgb2gray(wahaha1);
scenceImage = imread('scence.jpg');
sceneImage=rgb2gray(scenceImage);

第二步：检测特征点；

boxPoints = detectSURFFeatures(wahaha);
scenePoints = detectSURFFeatures(sceneImage);

figure; 
imshow(wahaha);
title('100 Strongest Feature Points from wahaha Image');
hold on;
plot(selectStrongest(boxPoints, 100));

figure; 
imshow(sceneImage);
title('300 Strongest Feature Points from Scene Image');
hold on;
plot(selectStrongest(scenePoints, 300));

第三步：提取特征描述子；

boxPairs = matchFeatures(boxFeatures, sceneFeatures);

第四步：显示假定匹配的特征；

matchedBoxPoints = boxPoints(boxPairs(:, 1), :);
matchedScenePoints = scenePoints(boxPairs(:, 2), :);
figure;
showMatchedFeatures(wahaha, sceneImage, matchedBoxPoints, ...
    matchedScenePoints, 'montage');
title('Putatively Matched Points (Including Outliers)');

第五步：使用场景中假定匹配定位物体；

[tform, inlierBoxPoints, inlierScenePoints] = ...
    estimateGeometricTransform(matchedBoxPoints, matchedScenePoints, 'affine');

figure;
showMatchedFeatures(wahaha, sceneImage, inlierBoxPoints, ...
    inlierScenePoints, 'montage');
title('Matched Points (Inliers Only)');

第六步：绘制定位框；

boxPolygon = [1, 1;...                           % top-left
        size(wahaha, 2), 1;...                 % top-right
        size(wahaha, 2), size(wahaha, 1);... % bottom-right
        1, size(wahaha, 1);...                 % bottom-left
        1, 1];                   % top-left again to close the polygon

newBoxPolygon = transformPointsForward(tform, boxPolygon);    

figure;
imshow(sceneImage);
hold on;
line(newBoxPolygon(:, 1), newBoxPolygon(:, 2), 'Color', 'y');
title('Detected Box');

第七步：检测另外一个目标；

cup1Image = imread('cup.jpg');
cupImage=rgb2gray(cup1Image);

cupPoints = detectSURFFeatures(cupImage);
figure;
imshow(cupImage);
hold on;
plot(selectStrongest(cupPoints, 100));
title('100 Strongest Feature Points from cup Image');

% Extract feature descriptors.
[cupFeatures, cupPoints] = extractFeatures(cupImage, cupPoints);

% Match Features
cupPairs = matchFeatures(cupFeatures, sceneFeatures, 'MaxRatio', 0.9);

% Display putatively matched features.
matchedcupPoints = cupPoints(cupPairs(:, 1), :);
matchedScenePoints = scenePoints(cupPairs(:, 2), :);
figure;
showMatchedFeatures(cupImage, sceneImage, matchedcupPoints, ...
    matchedScenePoints, 'montage');
title('Putatively Matched Points (Including Outliers)');

% Estimate Geometric Transformation and Eliminate Outliers
[tform, inlierElephantPoints, inlierScenePoints] = ...
    estimateGeometricTransform(matchedcupPoints, matchedScenePoints, 'affine');
figure;
showMatchedFeatures(cupImage, sceneImage, inlierElephantPoints, ...
    inlierScenePoints, 'montage');
title('Matched Points (Inliers Only)');

% Display Both Objects
cupPolygon = [1, 1;...                                 % top-left
        size(cupImage, 2), 1;...                       % top-right
        size(cupImage, 2), size(cupImage, 1);...  % bottom-right
        1, size(cupImage, 1);...                       % bottom-left
        1,1];                         % top-left again to close the polygon
 
newcupPolygon = transformPointsForward(tform, cupPolygon);    

figure;
imshow(sceneImage);
hold on;
line(newBoxPolygon(:, 1), newBoxPolygon(:, 2), 'Color', 'y');
line(newcupPolygon(:, 1), newcupPolygon(:, 2), 'Color', 'g');
title('Detected cup and Box');

displayEndOfDemoMessage(mfilename)

MATLAB特征点的检测与提取（2）

猜你喜欢