analysis of runner biomechanics using edge detection and image processing techniques to determine...
TRANSCRIPT
Analysis of Runner Biomechanics
Using Edge Detection and Image Processing Techniques to Determine Pronation Levels
Asa Kusuma
The Problem: Determining Pronation Pronation: Ankle rolling inward on impact Over-pronation major cause of running injuries Hard to determine with the eye, subtle
Pronation and Runners Every runner should pronate a moderate amount Too much pronation
Strains achilies tendon Puts pressure on Tibia Strains inner knee
Not enough pronation Supination Puts pressure on Fibula Strains IT Band
OverpronationSupination
The Setup
2nd Quarter
How the Setup Works Camera
Takes 20 Frames per second Produces 2 relevant images
Before impact At impact
Light Ensures universal lighting
Horizontal Tape Keeps runner in the same position relative to the camera
White Background Helps with edge detection
Method: Edge Shift
Get 2 Images, before and after impact Edge Detect Isolate the inner edge of foot/leg Find average x coordinate of edge
Equalize image edges Compare x coordinates of edges
Eliminating Noise
Upper portion of edge detection usually accurate Run normal edge detection, store edge pixels From top of edge, work down Only include edge pixels which have similar x-
coordinates Allow for more variance in x-coordinate for larger
differences in y-coordinate
Standardize images
Standard Not standard
Implementation of a GUI
Graphical User Interface library EasyGUI Allows event based GUI elements Runs within the python script Still must start program in terminal
Select Images
Select Which Leg
Select Pronation
Display Results
What Edges Look Like
NeutralSupinator Over-Pronator
Determining Relative Factors
Camera What’s the resolution of the camera? What’s the FPS of the image capture After testing, doesn’t matter
Shift Method How much shifting constitutes over-pronation? Testing on multiple subjects
1-5 pixel shift, supinator 6-10 pixel shift, neutral 11+ pixel shift, over-pronator