Published June 7, 2020 © GPL3+

Real-Time Object Tracking Using MATLAB (Blob Analysis)

A machine vision-based blob analysis method is explained to track an object in real-time using MATLAB and webcam.

BeginnerProtip1 hour6,275

Real-Time Object Tracking Using MATLAB (Blob Analysis)

Things used in this project

Software apps and online services

MATLAB

Support Package: OS Generic Video Interface

Story

Introduction

A machine vision-based blob analysis method is discussed in detail. The centroid value of an object is calculated from the image captured. The centroid value represents the location of the object in the workspace with respect to the image frame. In this work, red and blue objects are identified from the workspace.

Blob Analysis

Blob Analysis is a fundamental technique of machine vision based on analysis of consistent image regions. This tool is a choice for applications in which the objects being inspected are clearly discernible from the background. The main advantage of this technique to analyse an image includes high flexibility and excellent performance.

A blob is a group of connected pixels in an image that shares some common property (grayscale value). An example of Single blob and Multiple blob is shown in the below image. The white connected regions are blobs. The goal of blob detection is to identify and mark these regions.

(Left) Single blob (Right) Multiple blobs

The basic scenario of the Blob Analysis solution consists of the following steps:

1. Extraction – The RGB image is obtained as shown and it is converted into a Grayscale image with a threshold value. The threshold (0 to 1) is applied to obtain a region corresponding to the objects (or single object) being inspected as shown. In this example, red coloured objects are going to be detected.

(Left) RGB Image (Right) Grayscale Image

2. Refinement – The extracted region is often flawed by the noise of various kind (due to inconsistent lighting or poor image quality). In the refinement step, the image is enhanced by applying a noise filter (median filter). The median filter is a non-linear digital technique used to remove noise from an image. Fig. 3.5 shows the output after applying the filter. After applying the noise filter, the image is converted into a black and white image with a red threshold. Fig. 3.6 shows the output with only red components.

(Left) Noise Removed Image (Right) Black & White Image

3. Analysis – In the final step, the refined image is converted into a binary image and the final results are computed. If the image contains multiple objects, it is split into individual blobs each of which is inspected separately.

A bounding box is drawn over the blob. A Bounding Box of a blob is the minimum rectangle which contains the blob as shown. It is defined by finding the four pixels with minimum x-value, maximum x-value, minimum y-value and maximum y-value. From these values the width of the bounding box is given as xmax – xmin and the height as ymax – min.

(Left) Binary Image (Right) Blobs with Bounding box

Centroid Calculation

Once the blobs are detected and the bounding box is drawn over it, the centre of the blob (object) must be calculated as it represents the location of the object in the workspace.

The centroid (centre of mass) of a physical object is the location on the object where you should place your finger in order to balance the object. It is the average x- and y- location of the binary object. It is defined as a point, whose x-value is calculated by summing the x-coordinates of all pixels in the blob and then dividing by the total number of pixels. Similarly for the y-value. Mathematically, the centroid (x, y) of a blob (object) is calculated as in the below equation.

Centroid Calculation Formula

where N is the number of pixels in the blob. xi – yi is the x and y coordinates of the pixels respectively.

Software

In this work, Matlab 2016a is used. OS Generic Video Interface hardware Support Package must also be installed.

In order to check whether the camera device, either your inbuilt webcam of the laptop or your externally connected camera is configured in Matlab, type the following statement in the command window and hit enter.

>> imaqhwinfo

The imaqhwinfo function returns information about all image acquisition adaptors available on the system. The output should be something like shown below.

>> imaqhwinfo
ans =
InstalledAdaptors: {'winvideo'}
MATLABVersion: '9.0 (R2016a)'
ToolboxName: 'Image Acquisition Toolbox'
ToolboxVersion: '5.0 (R2016a)'

Here, winvideo is the inbuilt webcam of the laptop. That's it. All set to go! Copy-paste the code from the Code Section and Run the same in Matlab,

MATLAB Output

Real-Time Object Tracking using MATLAB

%%*********************************************************************
%%  Real-Time Object Tracking using MATLAB (Blob Analysis)
%%  Avinash.B (avinash15101994@gmail.com)
%%  https://avinashbaranitharan.com/
%%*********************************************************************

%% Initialization
redThresh = 0.10; % Threshold for red detection
vidDevice = imaq.VideoDevice('winvideo', 1, 'RGB24_640x480', ... % Acquire input video stream
                    'ROI', [1 1 640 480], ...
                    'ReturnedColorSpace', 'rgb');
vidInfo = imaqhwinfo(vidDevice);  % Acquire input video property
hblob = vision.BlobAnalysis('AreaOutputPort', false, ... % Set blob analysis handling
                                'CentroidOutputPort', true, ... 
                                'BoundingBoxOutputPort', true', ...
                                'MinimumBlobArea', 800, ...
                                'MaximumBlobArea', 3000, ...
                                'MaximumCount', 10);
hshapeinsRedBox = vision.ShapeInserter( 'BorderColor', 'Custom', ... % Set Red box handling
                                        'CustomBorderColor', [1 0 0], ...
                                        'Fill', true, ...
                                        'FillColor', 'Custom', ...
                                        'CustomFillColor', [1 0 0], ...
                                        'Opacity', 0.2);
htextins = vision.TextInserter('Text', 'Number of Red Object: %2d', ... % Set text for number of blobs
                                       'Location',  [7 2], ... 
                                       'Color', [1 0 0], ... % Red Colour
                                       'FontSize', 12);
htextinsCent = vision.TextInserter('Text', '+      X:%4d, Y:%4d', ... % set text for centroid
                                    'LocationSource', 'Input port', ...
                                    'Color', [1 1 0], ... // yellow color
                                    'FontSize', 14);
hVideoIn = vision.VideoPlayer('Name', 'Object Tracking - Avinash.B [2017209043]', ... % Output Video Player
                              'Position', [100 100 640 480]);
nFrame = 0; % Frame number initialization

%% Processing Loop
while(nFrame < 2000)
    rgbFrame = step(vidDevice); % Acquire single frame
    rgbFrameMirror = flipdim(rgbFrame,2); % Obtain the mirror image for displaying
    rgbgray=rgb2gray(rgbFrameMirror);
    diffFrameSub = imsubtract(rgbFrameMirror(:,:,1), rgbgray); % Get red component of the image
    diffFrame = medfilt2(diffFrameSub, [3 3]); % Filter out the noise by using median filter
    binFrame = im2bw(diffFrame, redThresh); % Convert the image into binary image with the red objects as white
    [centroid, bbox] = step(hblob, binFrame); % Get the centroids and bounding boxes of the blobs
    centroid = uint16(centroid); % Convert the centroids into Integer for further steps 
    rgbFrameMirror(1:20,1:165,:) = 0; % put a black region on the output stream
    vidIn = step(hshapeinsRedBox, rgbFrameMirror, bbox); % Instert the red box
    
    for object = 1:1:length(bbox(:,1)) % Write the corresponding centroids
        centX = centroid(object,1); %Store X Value in centX
        centY = centroid(object,2); %Store Y Value in centY
        vidIn = step(htextinsCent, vidIn, [centX centY], [centX-6 centY-9]); % Video Insert X,Y Values
    end
    
    vidIn = step(htextins, vidIn, uint8(length(bbox(:,1)))); % Count the number of blobs
    step(hVideoIn, vidIn); % Output video stream
    nFrame = nFrame+1;
end

%% Clearing Memory
release(hVideoIn); % Release all memory and buffer used
release(vidDevice);
clear all;
clc;

Credits

Avinash Baranitharan

2 projects • 13 followers

Master of the Possible, an Engineer.

Contact

Comments

Please log in or sign up to comment.

Embed the widget on your own site

Real-Time Object Tracking Using MATLAB (Blob Analysis)

Real-Time Object Tracking Using MATLAB (Blob Analysis)

Things used in this project

Software apps and online services

Story

Introduction

Blob Analysis

Centroid Calculation

Software

Code