Kinect-Based Gait Anomalies Detection

A person’s manner of walking, or their gait is an important feature for human recognition and classification tasks. Gait serves as an unobtrusive biometric modality, which yields high quality results. In comparison with other biometric modalities, its main strength is its performance even in data that is capturing at distance or at low resolution.The gait identification and classification is an important research direction in the field of vision. Depending on the input data, it can be mainly divide into two types:Video-based gait recognition and Kinect-based gait recognition. Kinect offers an attractive processing platform due to its low-cost build, non-intrusive acquisition, so we use this method.In this paper, we present an algorithm for classification of gait disorders arising from neurodegenerative diseases such as Parkinson and Hemiplegia. We focus on motion anomalies such as tremor, partial paralysis, gestural rigidity and postural instability. The analysis and classification of such motions is challenging since they consist a multiplicity of subtle formations while lacking a regular pattern or major cycle. We introduce a gait representation, which is invariant to the walking cycle and yields an efficient similarity metric.Our method performs on the 3D human skeleton and trajectories of joint points captured by a Kinect sensor. The algorithm is robust and does not require calibration, synchronization or a careful capturing setup. The experiments are conducting on real world datasets of different degenerative diseases. The results on all datasets demonstrate that our proposed method outperforms the baseline methods.Hence, our work makes the following contributions1. Introducing a non-intrusive, low-cost gait analysis algorithm, which allows for at-home, self-evaluation and examination of gait disturbances.2. We consider the full body information for gait classification. Thus, we calculate for all joints of the 3D skeleton generated from Kinect. This allows us to go beyond standard foot stride parameters yielding a rich set of features and improve accuracy.3. We define a dissimilarity measures between the test and trained gait models. It can recognize and delete redundant information and enhancing training data quality...