Activity 17: Basic Video Processing

In this activity, we are to analyze a video clip. The video clip we will analyze is a simple kinematic experiment and we are to extract certain physical parameters of the experiment and confirm the results analytically. To analyze the video, we need to separate its frames then on this frames we can apply the image processing techniques we have learned so far in the course. The experiment we designed is a simple rolling disk on an inclined plane. An animated gif of the set of images is shown below (video was cropped before images were extracted).


Our goal is to extract the acceleration of the object along the incline. First, we need to segregate the region of interest (ROI) from the background. From our previous lessons, we have two choices of image segmentation, color and binary. Since the video does not really contain important color information, I chose to perform binary image segmentation. The crucial step in this method is the choice of the threshold value. For this activity I found that a threshold value of 0.6 produced good results. However, the a good segmentation was still not achieved since the reflection of light in some parts of the image caused the color of the ramp to be unequal. A gif image of the result after thresholding is shown below:


Clearly, this is not what we want so I performed opening operation using a 10x5 structuring element both to further segregate the ROI from the background of the image and to make the edges of the ROI smooth. The resulting set of images is shown below:



Now we can track the position of the ROI in time. The position over time in pixels can be obtained from the average value of the x pixel position per image. Distance per unit time where distance is in pixels and time is in frames. The conversion factor we obtained is 2mm/pixel and from the video the frame rate is 24.5fps. After conversion I was able to obtain 0.5428m/sec^2.
Analytically, from Tipler, the acceleration along an inclined plane of a hollow cylinder is given by a = (g*sin(theta))/2, where theta is the angle of the incline, for our case theta = 6degrees. Using the formula we get 0.5121m/sec^2. This translates to roughly 5.66% of error. The scilab code I used is given below:

------------------------------------------------------------------------------
chdir("C:\Users\RAFAEL JACULBIA\Documents\subjects\186\activity17\hollow");
I = [];
x = [];
y = [];
vs=[];
ys = [];
xs = [];
as=[];
se = ones(10,5);
se2 = im2bw(se2,0.5);
for i = 6:31
I = imread("hollow" + string(i-1) + ".jpg");
I = im2bw(I,0.6);
I = dilate(erode(I,se),se);
imwrite(I , string(i)+ ".png" );
I = imread( string(i) + ".png");
[x,y] = find(I==1);
xs(i) = mean(x);
ys(i) = mean(y);
end
ys = ys*2*10**(-3);
vs = (ys(2:25)-ys(1:24))*24.5;
as = (vs(2:24)-vs(1:23))*24.5;
-------------------------------------------------------------

I grade myself 9/10 for this activity for not getting a fairly high error percentage.
Collaborators:
Ed David
JM Presto