The Research Of Footprints Extraction And Dynamic Recognition Based On Haptic Force Information

The haptic force information including footprint information, dynamic plantar pressure distribution information and so on, which contains the gait cycle, double support time, swing time, speed and other physiological, kinematics and dynamics characteristics. Those characteristics are widely used in many fields, such as diagnostic, gait recognition, gait analysis and sports analysis. In order to acquire the plantar pressure data, four acquisition methods such as pedography, sole barograph, insoles and force platform are used. The large-area flexible force-sensitive sensors developed by Intelligent Machines Institute of Chinese Academy of Sciences has a higher dot density, high precision and high frequency response characteristics, and we design and build the digital ground system based on the large-area flexible force-sensitive sensors. With the platform of the digital ground, we carry out some research about Parkinson gait analyse and aerobics choreography. But single step footprint data accurate segmentation and dynamic identification are the key technology and bottlenecks to carry out research. Therefore, a novel method for footprints extraction and dynamic recognition based on haptic force information is presented in this study. In this thesis we introduced the details to process the plantar pressure data, including the plantar pressure data filtering, single footprint plantar pressure data extraction and single footprint dynamic recognition. The main works in this thesis are introduced as follows:1. The method for footprints plantar pressure data extraction and dynamic recognition based on large-area flexible force-sensitive sensors. The operation includes the plantar pressure data filtering, single footprint plantar pressure data extraction and dynamic recognition. On account of the principle of plantar kinetics, the connected component algorithm is applied to data clustering analysis. Footprints recognition is accomplished on the basis of the foot anatomy principle and the appearance shape features. The experiments demonstrate that the method of footprints plantar pressure data extraction proposed in this thesis has a high recognition rate (99%) in the process of normal walking, and the recognition rate of single footprint dynamic recognition has achieved 97.5%. Aiming at the single footprint extraction accuracy of aerobics, the accuracy rate reaches 95%.2. The calculation of gait feature parameters and the establishment of gait cycle model and displacement model. Gait characteristics include temporal gait parameters, spatial gait parameters, plantar pressure distribution and so on. The gait cycle model and displacement model are used for track the footprints and calculate speed.3. The application of single footprint plantar pressure data extraction and dynamic recognition. To evaluate the gait balance, three gait features such as gait symmetry, gait static balance and dynamic balance were used. And we use CV-SVM to distinguish between the normal and abnormal gait. The experiment result shows that the gait balance parameters which are proposed in this thesis can reflect the difference between the normal and abnormal gait obviously, and can be used to distinguish the normal and abnormal gait.