Filtering Techniques Comparison in Gait Analysis using Depth Cameras

During the gait analysis using depth cameras, a filtering process is applied before procesing the signals, in order to reduce noise, fill null data and improve the gait parameters extraction. Such filtering process may alter the signal in a non desirable way and we consider important to verify if the processing we are using is adequate. It means that reduces enough noise without altering the esential parameters of the signal. For such task, we use two algorithms and compare both, in order to decide which one is the most reliable for the gait variables extraction with high confidence. In this paper, we compare both methods (interpolation and mean average) using a Pearson correlation and continuous concordance test between the captured signals before filtering and after filtering for each filter type. We found that both filtering methods produce similar results and are reliable.