Data Analysis, Variable Relationships, and Signal Processing… and then?

The graphs and tables below comprise the initial efforts at distilling the second data gathering sessions into a set of quantitative relationships between variables.  The tables reveal the most basic set of rules attained by comparing semi filtered data into relationships that were qualitatively determined by examining the data in graph form and performing peak finding, integral, smoothing, and derivative analysis on certain sets of data.

Much research was done into the theories and common practices within the biomedical biomechanics disciplines regarding the signal processing of EMG data to attain results.  This research will be mentioned here and presented in a more detailed fashion in a later post with complete references and equations used.

Experimentation so far has used moving window averaging and peak detection methods on EMG and PRESSURE data.  A kalman filter is used to integrate the gyro and accelerometer data into an error checked angle.  Further testing of EMG signal processing must be done as mentioned previously:  Root mean square and normalization can be used, as well as wavelet and FFT algorithms.  These may prove to be the most useful transformations providing frequency splitting of the data, and since EMG is usually bimodal, (active, and rest) these methods could prove useful, but will take longer to implement.  As my current goal is to identify relationships within the data variables with the larger goal of classifying the current behavior, drawing style, or action of the user it could be a fruitful method to break the states or classes of each variable into mini functions with different coefficients, this is the basis of walvelet transformations.

The pseudocode listed beside each graph quantifies a testable relationship that was observed within the dataset. The relationship reveals an action that was taken by the user. There is no psycoanalysis or inference of intention of the user within the data. I attempted to act only on the data and make a judgment only referring between the recorded video log and the data at the same time interval.

The tests that were written in pseudocode are also relationships that occurred more than once in the dataset. It is unclear how exclusive these relationships are. Does drawing in one way and enabling one test to return TRUE preclude the outcome of another test, this will be checked in further realtime tests implementing some form of feedback based on these logical tests.

This entry was posted in Thesis. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s