Sistema para Análise da Marcha Humana Baseado em Múltiplos Sensores Kinect v2

• Ano de defesa: 2018
• Autor(a) principal: Carvalho, Thiago Loureiro
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Engenharia Elétrica; Centro Tecnológico; UFES; Programa de Pós-Graduação em Engenharia Elétrica
• Palavras-chave em Português: Gait analysis; Data fusion; Multi-cameras; Rede de câmeras; Análise da marcha; Fusão de dados; Marcha humana - Análise; Multisensor para fusão de dados; Kinect (Controlador programável); Eletrônica Industrial, Sistemas e Controles Eletrônicos; 621.3
• Link de acesso: http://repositorio.ufes.br/handle/10/9571

Diseases like Parkinson, multiple sclerosis and rheumatoid arthritiscan affect some characteristics of people's normal gait, causingmobility problems that interfere with theirquality of life. The use of gait analysis has applications in sports, physiotherapy, and medicineand is very important for the diagnosis of some diseases. Gait analysis tools can be very useful in the treatment of various diseases and inmotor rehabilitation of patients. However, the high cost of commercial tools for gait analysis makesit unfeasible for use in small clinics. This work proposes an RGBd (Kinect v2) multi-camera system to identify the position of the human’s center of mass andjoints for gait analysis. Kinect v2 also has advantages over other computer vision systems used for gait analysis: low cost and no need to use reflective markers on the human's body. However, the use of several sensors brings the need for synchronization and extrinsic calibration of the camera network, in addition to data fusion. Yet, the capture area increases, problems of occlusion are avoided and the precision in the measurement of parameterscan be improved.From the data provided by the system, spatio-temporal parameters of the gait can be calculated: step length, stridelength, gait speed and cadence.Preliminary tests indicate that the data fusion with the Kalman filter works well. The validation experiments of the precision for calculation of the parameters by the system show that the estimated parameters are coherent with the expected ones. It was also possible to perceive a consistency in the estimates, because, the standard deviation is small in all the experiments. The participant’s free gait parameters were also calculated, and the results are compatible with the results found in the literature. The calculated steps using the proposed system havemean of0,66 m in length with a standard deviation of 0,04 m for an average speed of 1,04 m/s, very close to the results found in the literature (0,7 m, with a standard deviation of 0,06 m, for an average velocity of 1,2 m/s).With the developed system waspossible to calculatethe temporary space parametersconsistently and coherentlyfrom three low-cost sensors (Kinect v2) without using markers on the user's body.