Gait analysis is a procedure applicable not exclusively in sports science or basic biomechanical research, but also in clinical diagnosis, musculoskeletal rehabilitation and functional recovery monitoring. In the clinical practice, human gait can be analysed without the use of any instrumentation – by means of visual examination – or using optoelectronic systems, tensometric platforms, wearable sensors, or other types of devices. In recent years, there has been an increasing number of research projects being carried out in order to develop gait analysis systems based on depth sensors; however, such systems are not yet commercially available. The depth sensor is relatively inexpensive, portable, and simple in design. The availability of algorithms for detecting human silhouettes in depth images and for estimating the coordinates of selected body points makes depth sensors particularly suitable for human gait analysis applications.

This thesis is focused on the estimation of spatio-temporal parameters characterising human gait, based on depth sensor data, and also on the comparison of the estimates obtained using those methods, with results obtained by using the Zebris–FDM gait analysis system. This thesis includes a review of literature related to gait analysis and the techniques used for it, the acquisition of data by completing a series of experiments and the implementation of three data processing methods in the MATLAB R2022a environment. Method 1 is based on the analysis of the anterior-posterior distance between the knees and involves the determination of the local extrema of that distance, which represent the moments of placing the feet on the ground. Method 2 is based on the analysis of the spine base height and involves the determination of the intervals between the local maxima of that height, which allows the length and time of the steps to be estimated. Method 3 is based on the analysis of the instantaneous velocity of the feet and involves the numerical differentiation of the time courses of the positions of feet and the determination of the moments in which feet velocities fall below or rise above an empirically selected threshold. Those moments correspond to placing feet on the ground or detaching them from it. The criteria used in this thesis for verifying the effectiveness of the methods are: mean error, standard deviation, and the mean absolute value of the relative error.

Each method yields estimates of temporal gait parameters close to the reference values. For spatial parameters, Method 3 is the most accurate. The results of the reported research also allow for the identification of practical advantages and disadvantages of the studied data-processing methods, as well as opportunities for their further development.