SURF features for object tracking

Hi!

I am asking a question related to a previous question I asked: https://es.mathworks.com/matlabcentral/answers/1984899-single-object-tracking-using-sift-features

In there you can find the origin of my problem. So I am using SURF features now instead of SIFT features for matching a motorcycle to a set of frames where this motorcycle appears. The thing is that no matter what I try, there is always a frame that gives the following error in line T = estimateGeometricTransform2D(m_kp_obj, m_kp_esc, "affine") :

Error using vision.internal.geotrans.algEstimateGeometricTransform>checkRuntimeStatus Could not find enough inliers in matchedPoints1 and matchedPoints2. Error in vision.internal.geotrans.algEstimateGeometricTransform (line 72) checkRuntimeStatus(statusCode, status, sampleSize); Error in estimateGeometricTransform2D (line 157) vision.internal.geotrans.algEstimateGeometricTransform(...

I have changed descriptors from SIFT to SURF, changes MatchThreshold value, change the points required for matching, but it still gives me this error and is a completely nightmare. I hope someone really could help me fixing this. I now provide the code, please feel free to use it and test it, if that means understanding better my issue or fixing it. If interested, you can find the link to all frames folder in the previous related questoin I made (link above).

Thank you very much. I really hope someone can help me. I'm open to do discord or whatever if that means making things easier.

Here is the code:

% myTracker, codi inicial del short project

close all

% Llegim el fitxer d'anotacions groundtruth_rect.txt: frame, bounding boxes, is_lost

BB = importdata('./MotorcycleChase/groundtruth_rect.txt');

Idir = dir('./MotorcycleChase/img/*.jpg');

% figure

% hold on % mantenim sempre oberta la mateixa figura

filename = horzcat(Idir(1).folder,'/',Idir(1).name);

I = imread(filename);

imshow(I);

moto = imcrop(I,BB(1,2:5));

imshow(moto);

im_obj = rgb2gray(moto);

kp_obj = detectSURFFeatures(im_obj);

kp_obj = selectStrongest(kp_obj,50);

[feat_obj,kp_obj] = extractFeatures(im_obj,kp_obj);

nf = size(Idir);

for i = 2:nf

filename = horzcat(Idir(i).folder,'/',Idir(i).name);

im_esc = rgb2gray(imread(filename));

kp_esc = detectSURFFeatures(im_esc);

kp_esc = selectStrongest(kp_esc,50);

[feat_esc,kp_esc] = extractFeatures(im_esc,kp_esc);

pairs = matchFeatures(feat_obj, feat_esc, 'MatchThreshold', 3);

m_kp_obj = kp_obj(pairs(:, 1), :);

m_kp_esc = kp_esc(pairs(:, 2), :);

if size(m_kp_obj, 1) >= 3 && size(m_kp_esc, 1) >= 3

T = estimateGeometricTransform2D(m_kp_obj, m_kp_esc, "affine");

[f, c] = size(im_obj);

box = [1, 1; c, 1; c, f; 1, f; 1, 1];

nbox = transformPointsForward(T, box);

nbox = sortrows(nbox, [2 1]); % Sort points based on y, then x

% Calculate the new bounding box coordinates (nbox2)

minCoord = min(nbox);

maxCoord = max(nbox);

nbox2 = [minCoord(1), minCoord(2), maxCoord(1)-minCoord(1), maxCoord(2)-minCoord(2)];

overlapRatio = bboxOverlapRatio(nbox2, BB(i, 2:5));

imshow(im_esc); % Show the frame

rectangle('Position', nbox2, 'EdgeColor', 'blue');

drawnow

else

% Handle the case when there are not enough matched points

% You can either skip this frame or use a different approach

end

end

Answer 1

[Ranjeet]

Hi Pol,

The key message in the provided error message is “Could not find enough inliers in matchedPoint1 and matchedPoint2”.

Have a look at the following resources discussing the same issue and proposed debugging/solution –

1. Error: Could not find enough inliers in imagePoints and worldPoints
2. estimateGeometricTransform don’t return the status

Also, ‘estimategeometricTransform’ is not recommended to use, you may refer to the following alternatives –

1. estgeotfrom2d
2. estgeotfrom3d