Live Image Processing and Tracking Image Features

12 次查看（过去 30 天）
Scott Kaiser
Scott Kaiser 2023-8-29
回答： Image Analyst 2023-8-29
I am attempting to adopt the functionality of the 'Fluorescence-Tracker-App' for use with a Mako G-507B camera instead of a video. 'Fluorescence-Tracker-App' can be downloaded from mathworks but I am trying to change it so it works with the camera I have. My idea is to just take a snapshot, select a region of interest, find 4 features, and track those features in the following images that are captured. The code I have is shown below. The problem I am running in to is that the camera times out and is unreliable, it seems, in updating and/or even taking snapshots. So the program doesnt work every time. I would like to do this processing live, and as quickly as is reasonable for the processing to take place. Does anyone have any suggestions to make this program run more reliably?
clear;clc;close all;
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %
% Parameters for Camera
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %
g= gigecam;
exposureTime=140;%[us]
g.Timeout=80;
g.ExposureAuto = 'Off';
g.ExposureMode = 'Timed';
g.ExposureTimeAbs = exposureTime;
g.PixelFormat='Mono12';
g.Height=484;
g.Width=484;
N=20; %Total Number of images to take.
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %
% Select and add regions of interest
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %
region = [];
img=snapshot(g);
marked = mat2gray(img);
fig = figure(Name="Double-click to accept. " ...
+"Close figure when complete.", NumberTitle="off");
while isvalid(fig)
[~,roi] = imcrop(marked);
region = [region; roi]; %#ok<AGROW>
marked = insertShape(marked, Rectangle=roi, Color="white");
end
% If closed without a selection, set default regions [x y w h]
if isempty(region)
region = [400 20 40 60; 180 220 60 60; 220 40 60 40];
end
% Draw the selected region(s) of interest
marked = insertShape(mat2gray(img), Rectangle=region, Color="white");
marked = insertShape(marked, Rectangle=region+[0 double(g.Height) 0 0], Color="white");
figure; imshow(marked);
title("Selected Region(s) of Interest");
maxPts = 4;
points = []; rid = [];
gframe = im2gray(mat2gray(img));
for j = 1:size(region,1)
pts = detectMinEigenFeatures(gframe, ROI=region(j,:));
pts = selectStrongest(pts,maxPts);
points = [points; pts.Location];
rid = [rid; j*ones(size(pts,1),1)];
end
% Use region id (rid) to color code points
cmap = lines(max(rid));
cmap = cmap(rid,:);
% Display the detected points
marked = insertMarker(marked,points,"+", Color=255*cmap);
marked = insertMarker(marked,points+[0 double(g.Height)],"+", Color=255*cmap);
figure; imshow(marked)
title("Detected Features");
tracker = vision.PointTracker(MaxBidirectionalError=1);
initialize(tracker,points,mat2gray(img));
player = vision.VideoPlayer(Position=[60 60 460 610]);
k = 1;
while k<=10*N
k = k + 1;
% Get the next snapshot
frame = snapshot(g);
cframe = mat2gray(imcrop(frame,[0 0 double(g.Width) double(g.Height)]));
% Track the points, note some points may become lost/invalid
[points,val] = tracker(cframe);
if any(val)
% Store corresponding intensity values
idx = sub2ind([double(g.Height) double(g.Width)],round(points(val,2)),round(points(val,1)));
% Mark frames with tracked points
markerSize = 10; % Adjust the marker size as needed
lineWidth = 5; % Adjust the line width as needed
cframe = insertMarker(cframe, points(val,:), 'o', ...
'Color', [255, 0, 0], 'Size',markerSize);
end
% Display annotated video frame using the video player object
player((cframe))
end
% Clean up
release(player)
release(tracker)

请先登录，再回答此问题。

采纳的回答

Image Analyst
Image Analyst 2023-8-29
Call tech support and ask for Andrew B in the image acquisition team. He helped me out with such an issue. There are other functions to use other than snapshot such as start and getdata
data = getdata(obj) returns data, which contains the number of frames specified in the FramesPerTrigger property of the video input object obj. obj must be a 1-by-1 video input object.

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 GigE Vision Hardware 的更多信息

标签

产品


版本

R2022a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by