Track And Map

Uses

• OpenCV 2.4.6
• Matlab 8.2.0.701

Tracks video input using an HSV filter and outputs a MATLAB figure analyzing movement

Aquiring Data

In the terminal navigate through folders using commands:

• ls will list files and folders of the current directory
• cd 'folder name' will change into directory 'folder name'
• cd .. will navigate to the parent directory
• Right clicking on a file and selecting properties will give you a file's location

To convert files from .wmv to .avi

1. In the terminal navigate to the folder containing the files
2. Enter ~/Convert.sh and the program will convert all the files
3. After conversion it maybe necessary to crop out similar colors using Video Pad

To create the tracking data

1. In the terminal enter ObjRec /path/to/video.avi /desired/path/to/data.csv
2. Once application opens, reposition the four windows so that each window is visable and use the HSV sliders to isolate the color of interest in the threshold window (for neon pink bringing the S_MIN up to ~100, the V_MIN up to ~110 and the S_MAX down to ~150 worked well for me)
3. Once cross hair is consistently tracking, switch from pause to play and switch recording on
4. Wait for video playback to finish
5. Once recording is complete, open .csv folder to validate results

Graphing results in MATLAB

Importing

1. Select 'Import Data' in the toolbar
2. Navigate to and select the .csv file
3. Select Matrix instead of the default import type Column Vectors
4. 'Import Selection'
5. Rename the matrix to framexy by right clicking on the matrix in the workspace

Heatmap Graphing

1. Select 'New Script' from the toolbar and copy and paste the contents of heatmap.m into the script and save it as heatmap.m or clone heatmap.m from this repository into your working MATLAB directory
2. In the Command Window enter xy = horzcat(framexy(:,2), framexy(:,3))
3. heatmap(xy) will display a figure with the heatmap overlaid by the vector of motion

Overlaying Audio Cues (must have signal processing toolkit)

Video Frame to Time Conversion

1. frames = length(framexy)
2. time = frames/62.5
3. tv = linspace(0, time, frames)
4. tv = transpose(tv)
5. timexy = horzcat(tv(:,1), framexy(:,2), framexy(:,3))

Audio Extraction from .avi

1. Go to Audio Extractor
2. Upload .avi and choose .wav as the extraction output
3. Add the extracted audio to your Matlab project folder

Audio Processing

1. A Matlab project uses a current working directory which can be seen in the Current Folder pane
2. Put filename.wav in current working directory
3. In the Command Window enter [f, Fs] = wavread('filename.wav')
4. N = length(f)
5. slength = N/Fs
6. ta = linspace(0, slength, N)
7. ta= transpose(ta)
8. left = f(:,1)
9. wav = [ta, left]
10. plot(wav(:,1),wav(:,2)) in order to see what you would like to use as the upper cut off threshold
11. indices = find(wav(:,2) >= .3) .3 was the upper cut off threshold used here but this will vary, choose a value you feel appropriate from the previous step
12. wav(indices,:) = []

Heatmap with Audio Cues plotted

1. Select 'New Script' from the toolbar and copy and paste the contents of AVcompare.m into the script and save it as AVcompare.m or clone AVcompare.m from this repository into your working MATLAB directory
2. Enter heatmap(xy)
3. Enter AVcompare(timexy, wav)