Recently, with the development of machine learning theory to a wide range of innovative learning and classification algorithms with better performance has been more and more prevalent in the modern pattern recognition. to assess the dynamics of human gait function. For example, artificial neural network (ANN) based on machine learning algorithms have been employed for automatic recognition gait pattern changes using different walks of data [1,2. ] Holzreiter and Kohle used ANN to identify walking normal and pathological using kinetic data [1] pulp and Barton used ANN to distinguish gait patterns that vary according to the nature of the walk is extracted from the hip and knee. measures joint angles We know that walking algorithm classification performance classification mainly based on the property walk extract or choose from the default variable walk, that is, before processing for variable walking. Starting the first step for improving the classification performance [3,4. ]
we know that the traditional method of selection of the corridor features some of the parameters have been selected waveforms from a temporary property to walk the walk for the model. For example, the bell and the different values of incident happened during a walk around the property has been selected for the walk. In fact, the passage of the complex nature of the changes that are either non-stationarity and non-linearity and all the information about the manner of walking, working in collaboration on the move using a variable. It is essential that techniques to extract features from walking to treat both the time and frequency of walking. Unfortunately, the property had been walking by traditional methods can not provide important information relevant to the detection of the characteristic changes of the functions of walking. This encourages the application of advanced data analysis techniques to go for the capture of time-frequency information about the function of the human gait [3,4]
It is well known that analysis. Wave [5] is a powerful technique that can provide both spectral and time, simultaneous data has been successfully applied in biomedical signal processing, feature extraction [5,6] in this study. For the sake of improving the generalization performance classifier ANN. We addressed the following novel form of pattern-wise for the automatic recognition of the walking pattern. The wavelet transform is used first to extract some good properties of strongly correlated. timedependent Then walk the walk features a variable used to start a training set of Ann. To assess the pattern present data efficiently walking kinetics of 24 young and 24 elderly people were coming and analysis. In addition, we compared the proposed model of a pattern classification based on the traditional choice of walking the property
This paper is organized as follows: section II presents the processing steps. Data to walk using our proposed model. Discussion and conclusions are given in Section.
การแปล กรุณารอสักครู่..
